Server parameter configuration method, electronic equipment, storage medium and program product

Abstract

The invention discloses a server parameter configuration method, electronic equipment, a storage medium and a program product, and relates to the technical field of servers. The method comprises the following steps: constructing a public key-value according to each standard configuration parameter subclass of to-be-configured information of a server, determining a corresponding relationship between each source configuration item and a public key by identifying each source configuration item of a source server and a semantic word block, which is the same as the semantic word block of the public key in the public key-value, of the public key, based on the condition that the configuration contents of the configuration items with the same semantics are the same, the configuration contents of the source configuration items are mapped into values corresponding to the public keys, and finally the mapped public key-values are mapped to the to-be-processed configuration items of the target server needing parameter configuration again according to the same mode. According to the method and the device, the problem that server parameters cannot be efficiently and accurately configured in related technologies can be solved, and automatic and accurate copying of server configurations of cross-manufacturer models can be realized.

Description

Technical Field

[0001] The present invention relates to the technical field of servers, and in particular, to a server parameter configuration method, an electronic device, a computer-readable storage medium, and a computer program product. Background Art

[0002] A data center includes computing power devices represented by servers, and all servers need to be configured with the same configuration items during operation and maintenance. In the server parameter configuration method of the related art, due to the significant differences in the configurations and models of different servers, efficient and accurate parameter configuration cannot be achieved. Summary of the Invention

[0003] The present invention provides a server parameter configuration method, an electronic device, a computer-readable storage medium, and a computer program product, which realize automatic and accurate replication of server configurations across different manufacturers' models.

[0004] To solve the above technical problems, the present invention provides the following technical solutions: On the one hand, the present invention provides a server parameter configuration method, including: Redefine the to-be-configured information of the servers in the data center as multiple standard configuration parameter subclasses, and construct a common key-value based on each standard configuration parameter subclass, where each standard configuration parameter subclass corresponds to at least one configuration item.

[0005] Obtain each source configuration item of the source server with configured parameters, determine the corresponding relationship between each source configuration item and the common key by identifying the same semantic phrase blocks of the common key of each source configuration item and the common key-value, and map the configuration content of each source configuration item to the value corresponding to the common key based on the condition that the configuration contents of the configuration items with the same semantics are the same.

[0006] Determine the corresponding relationship between each to-be-processed configuration item and the common key by identifying the same semantic phrase blocks of the common key in the target server, and generate the corresponding configuration content for each to-be-processed configuration item based on the mapped common key-value to complete the parameter configuration of the target server.

[0007] The present invention also provides an electronic device, including a memory and a processor, and the processor is used to implement the steps of any one of the above server parameter configuration methods when executing the computer program stored in the memory.

[0008] The present invention also provides a computer-readable storage medium, which stores a computer program, and the computer program is used to implement the steps of any one of the above server parameter configuration methods when executed by a processor.

[0009] Finally, the present invention also provides a computer program product, including computer programs / instructions, which implement the steps of any of the above server parameter configuration methods when executed by a processor.

[0010] The advantages of the technical solution provided by the present invention are as follows: by identifying the semantic information between the configured items of the configured server and the standard configuration parameters defined by the common key, the problem of different names but the same meaning between the source server and the standard configuration parameters can be solved. Through the semantic recognition between the common key and the configured items of the target server, the manufacturer parameter differences are shielded. After double-layer mapping, the mapping relationship from the source server to the configured items of the target server can be determined, solving the problem of excessive semantic differences between the configured items from the source server to the target server. Through conversion with the common key, the accuracy and efficiency of mapping the source configured items to the target configured items are effectively improved. The current configuration information of the source server is converted into all configured items and copied to the required configuration parameters of the target server, realizing the automatic and accurate replication of the server configuration across different manufacturer models, minimizing the manual intervention in the parameter configuration process of the servers in the data center. It can not only greatly improve the efficiency and accuracy of configuring the servers, but also effectively avoid configuration failures or configuration deviations caused by manual operation errors, and effectively shorten the cycle from the preparation to the online operation of the servers.

[0011] In addition, the present invention also provides corresponding implementation electronic devices, computer-readable storage media, and computer program products for the server parameter configuration method, further making the method more practical, and the electronic devices, computer-readable storage media, and computer program products have corresponding advantages. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions of the present invention or related technologies, the following will briefly introduce the drawings required for use in the description of the embodiments or related technologies. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0013] Figure 1 It is a schematic diagram of the hardware composition framework applicable to the server parameter configuration method provided by the present invention; Figure 2 It is a schematic flowchart of a server parameter configuration method provided by the present invention; Figure 3 It is a schematic diagram of the corresponding relationship between the standard configuration parameter subclasses and their configured items in an exemplary application scenario provided by the present invention; Figure 4 It is a schematic diagram of two-layer mapping in an exemplary application scenario provided by the present invention; Figure 5Schematic flowchart of another server parameter configuration method provided by the present invention; Figure 6 Structural framework diagram under an exemplary embodiment of the server parameter configuration device provided by the present invention; Figure 7 Structural diagram of an exemplary embodiment of the electronic device provided by the present invention. Detailed implementation manners

[0014] In order to enable those skilled in the art to better understand the technical solutions of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific implementation manners. Among them, the terms "first", "second", etc. in the specification and the above drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. The term "exemplary" means "serving as an example, embodiment, or illustration". Any embodiment described herein as "exemplary" does not necessarily need to be construed as superior to or better than other embodiments.

[0015] With the rapid development of artificial intelligence, big data, and cloud technologies, in order to meet the exponentially growing data processing, data centers that can efficiently store and manage large-scale data and have high availability guarantees have emerged. A data center is a place that provides an operating environment for centrally placed electronic information devices, and centrally stores, calculates, and exchanges data. It is the underlying core infrastructure of cloud computing, which includes computing power (i.e., data processing capabilities) devices of information technology represented by servers, as well as basic support facilities to ensure the normal operation of these devices, such as power supply and distribution systems, refrigeration systems, etc. In the operation and maintenance of a data center, all servers need to be configured with the same configuration items. Since there are different models of servers from different manufacturers in the data center, and there are differences in the configurations of different servers, such as inconsistent parameter names, inconsistent structures, and ambiguous semantics, etc., in the related art, during the configuration process of the servers in the data center, experienced technical personnel still need to intervene, such as manually configuring certain parameters on the server management page, resulting in low efficiency, high error rate, and poor scalability in the configuration of the data center.

[0016] In view of this, in order to achieve the automated and accurate replication of server configurations across different manufacturers' models and solve the problems that manual intervention has high requirements for the experience of operators, takes a long time to configure, and has the risk of inaccurate configuration, the present invention proposes a two-layer mapping method: matching standard semantic tags between the configured items of the server and the common keys, and then mapping to the configured items of the target server using the standard tags, determining the mapping relationship of the configured items from the source server to the target server through the common keys, solving the problem of excessive semantic differences in the configured items between the source server and the target server, improving the accuracy of mapping the source configured items to the target configured items, and achieving the automated and accurate replication of server configurations across different manufacturers' models. Combining with the specific application environment architecture or specific hardware architecture on which the execution of the server parameter configuration method depends, the specific application environment architecture or specific hardware architecture is described herein. The following combines Figure 1 Some possible application scenarios related to the technical solution of the present invention are introduced by way of example. The data center includes multiple servers with exactly the same configuration. The configuration process of each server may include the following content: Randomly select a server in the data center as the source server 101, and use the other servers as the target servers 102. Redefine the information to be configured of the servers in the data center as multiple standard configuration parameter subclasses. Each standard configuration parameter subclass corresponds to at least one configured item, and construct a common key-value based on each standard configuration parameter subclass to perform server parameter configuration on the source server. The parameter configuration process for any one of the target servers is as follows: for each source configured item of the source server 101 with already configured parameters, determine the corresponding relationship between each source configured item and the common key by identifying the same semantic chunks of the common key of each source configured item and the common key-value, and based on the condition that the configuration contents of the configured items with the same semantics are the same, map the configuration contents of each source configured item to the value corresponding to the common key; determine the corresponding relationship between each to-be-processed configured item in the target server 102 and the common key by identifying the same semantic chunks of each to-be-processed configured item in the target server 102 and the common key, and generate the corresponding configuration content for each to-be-processed configured item based on the mapped common key-value to complete the parameter configuration of the target server 102.

[0017] It should be noted that the above application scenarios are only shown for the convenience of understanding the idea and principle of the present invention, and the embodiments of the present invention are not limited in this regard. On the contrary, the embodiments of the present invention can be applied to any applicable scenario. After introducing the technical solution of the present invention, the various non-limiting embodiments of the present invention will be described in detail below with reference to the drawings and specific embodiments. First, please refer to Figure 2 , Figure 2 is a schematic flowchart of a server parameter configuration method provided in this embodiment. This embodiment may include the following content: S201: Redefine the configuration information to be configured for the servers in the data center as multiple subclasses of standard configuration parameters, and construct common key-values based on each subclass of standard configuration parameters.

[0018] In this step, the configuration information to be configured refers to all configurations of the server configuration parameters in the business scenario of the current data center. For the convenience of subsequent mapping processing, all configurations can be redefined and summarized into multiple configuration subclasses. For the convenience of description, they are defined as subclasses of standard configuration parameters. For example, the configuration of each server can be summarized into configuration subclasses such as NTP (Network Time Protocol) configuration, service configuration, alarm configuration, user configuration, mailbox configuration, Domain Name System configuration, fan configuration, Lightweight Directory Access Protocol configuration, and log configuration. Among them, NTP is a configuration function of server configuration, which synchronizes the current time of the server by configuring one or more time servers. Each subclass of standard configuration parameters corresponds to at least one configuration item. Obtain the corresponding configuration items for each subclass of standard configuration parameters. As Figure 3 shown, convert each subclass of standard configuration parameters and its respective configuration items into the form of key-value. For the convenience of description, converting each subclass of standard configuration parameters and its respective configuration items into key-value is defined as a common key-value. Exemplarily, all configuration items of all subclasses of standard configuration parameters can be taken as a whole and then converted into a key, and the value can be empty or a default value is adopted. Each standard configuration parameter and configuration item can be described in any language. For example, it supports unified semantic mapping of multi-language configuration items such as Chinese ("Network Time Server") and English ("Network Time Server"). Through Unicode (Universal Code) encoding and chunk alignment technology, the accuracy of cross-language configuration replication is achieved.

[0019] S202: Obtain each source configuration item of the source server with parameters already configured. Determine the corresponding relationship between each source configuration item and the common key by identifying the same semantic chunks of each source configuration item and the common key. And based on the condition that the configuration contents of the configuration items with the same semantics are the same, map the configuration contents of each source configuration item to the value corresponding to the common key.

[0020] In this step, obtain the configuration information of the source server. The configuration information includes at least each configuration item and its corresponding configuration content. For the sake of distinction, the configuration items of the source server are defined as source configuration items. Source configuration items may include NTP configuration, service configuration, alarm configuration, log configuration, mailbox configuration, domain name system configuration, fan configuration, Lightweight Directory Access Protocol configuration, user configuration, etc. The obtained configuration information can also be converted into the key-value format, where the source configuration item serves as the key and the corresponding configuration content serves as the value. Among them, the source configuration item and the configuration content can be described in any language. For example, it supports unified semantic mapping of multi-language configuration items such as Chinese representation: "Network Time Server" and English representation: "Network Time Server". Through Unicode (Universal Code) encoding and chunk alignment technology, the accuracy of cross-language configuration replication is achieved.

[0021] It can be understood that due to different server manufacturers and different server models, the parameter names of the configuration items of the server are different. To shield the differences between the source server and the target server, the corresponding relationship between each source configuration item of the source server and each configuration item of each standard configuration parameter subclass can be established first. That is, identify the situation where the source configuration item in the source server and the configuration item of the standard configuration parameter subclass essentially belong to the same configuration item but use different configuration names. For accurate identification, the common key and each source configuration item can be identified according to the chunks. Any relevant technology capable of identifying the semantics of chunks can be used, which does not affect the implementation of the present invention. The chunks with the same semantics in this step refer to two chunks that essentially express the same meaning. The same meaning includes describing the same object in different languages and also includes using similar descriptions. After determining the corresponding relationship between each source configuration item and each configuration item of each standard configuration parameter subclass, the configuration content of the source configuration item with the corresponding relationship, that is, the value of the key-value pair corresponding to the source configuration item, is used as the value of the configuration item of the standard configuration parameter subclass corresponding to the common key, completing the process of mapping each source configuration item of the source server to the common key.

[0022] S203: By identifying the same semantic chunks between each to-be-processed configuration item in the target server and the common key, determine the corresponding relationship between each to-be-processed configuration item and the common key, and generate the corresponding configuration content for each to-be-processed configuration item based on the mapped common key-value to complete the parameter configuration of the target server.

[0023] Among them, both the target server and the source server are servers in the data center. The target server is the server that needs to configure server parameters according to the source server. When the above S201 completes extracting the keys of all types of servers as the common keys, and S202 completes mapping the configuration items of the source server to the common keys, the configuration items to be configured for the target server can be collected in the same way as S202. For the convenience of description, the configuration items to be configured for the target server are defined as the configuration items to be processed. The same identification method as the above steps is used to determine the correspondence between the standard configuration parameter subclasses and each configuration item to be processed, and the common key is mapped to the configuration items to be processed of the target server, as Figure 4 shown. The configuration items after the two mappings are used as the parameter content for configuring the configuration items to be processed of the target server. After double-layer mapping, the mapping relationship between the configuration items of the source server and the target server is determined, and the current configuration of the source server is converted into all configuration items and copied to the configuration parameters required by the target server. For example, if the corresponding value of the source server configuration item SERVER_NTP_IP1 is 192.168.1.1, and the result after mapping the configuration item is NTP_SERVER1, that is, SERVER_NTP_IP1 of the source server corresponds to NTP_SERVER1 of the target server, then NTP_SERVER1 in the configuration item to be processed of the target server can be configured as 192.168.1.1. After all the configuration items are mapped, the target server configuration item parameters are generated, and the configuration items to be processed of the target server can be configured accordingly by calling the Redfish (standard protocol name) interface.

[0024] In the technical solution provided in this embodiment, by identifying the semantic information between the configuration items of the configured server and the standard configuration parameters defined by the common key, the problem of different names but the same meaning between the source server and the standard configuration parameters can be solved. Through the semantic recognition between the common key and the configuration items of the target server, the manufacturer parameter differences are shielded. After double-layer mapping, the mapping relationship between the configuration items of the source server and the target server can be determined, and the problem of excessive semantic differences between the configuration items of the source server and the target server can be solved. Through conversion with the common key, the accuracy and efficiency of mapping the source configuration items to the target configuration items are effectively improved. The current configuration information of the source server is converted into all configuration items and copied to the configuration parameters required by the target server, realizing the automatic and accurate replication of the server configuration of cross-vendor models, and minimizing the manual intervention in the parameter configuration process of the servers in the data center. It can not only greatly improve the efficiency and accuracy of configuring the server, but also effectively avoid configuration failures or configuration deviations caused by manual operation errors, and effectively shorten the cycle from the preparation of the server to its online operation.

[0025] In the above embodiments, there is no limitation on how to collect the configuration information of each server in the data center. This embodiment also provides an exemplary implementation method, which may include the following: Data collection components can be pre-set on each server in the data center to call the data collection components to collect the configuration information to be configured of the servers in the data center, each source configuration item of the source server, and the configuration item to be processed of the target server. Among them, the data collection component at least encapsulates the Redfish API (Application Programming Interface) and scripts. Users can manage the server through the Redfish API based on the standardized RESTful (web API interface designed based on the REST (Representational State Transfer) architecture style) interface, and perform operations such as obtaining sensor data, configuration parameters, and executing power control. The script can be written in any script language to obtain relevant information of the baseboard management controller in cooperation with the Redfish client library.

[0026] As can be seen from the above, through Redfish, this embodiment can simplify the management of physical server resources, easily manage various resources of physical servers, improve the monitoring efficiency of data center servers, and obtain server configuration information efficiently and accurately.

[0027] In the above embodiments, there is no limitation on how to determine the correspondence between each source configuration item and the common key. This embodiment also provides an exemplary implementation method, which may include the following: As Figure 4 shown, each word block of each source configuration item and the common key is extracted respectively, and the corresponding source word block sequence and common word block sequence are generated; based on the word block position and the initial value of the word block, the semantic quantization values of each word block in each source word block sequence and common word block sequence are calculated respectively; the similarity degree between the semantic quantization values of each word block in each source word block sequence and the semantic quantization values of each word block in the common word block sequence is calculated, and two word blocks that meet the preset similarity condition are used as the same semantic word blocks.

[0028] Among them, the word chunks of each source configuration item and the common key can be extracted by a word segmentation method, and the extracted word chunks are combined into a sequence according to the extraction order. For the convenience of automatic detection, recognition identifiers can be added before and after the sequence. That is, exemplarily, the word segmentation process can be performed on each source configuration item to obtain a set of source configuration word chunks corresponding to each source configuration item, and the sequence start identifier and the sequence end identifier are added to each group of source configuration word chunks to obtain each source word chunk sequence; the word segmentation process is performed on the common key to obtain a set of common word chunks, and the sequence start identifier and the sequence end identifier are added to the common word chunks to obtain the common word chunk sequence. The word segmentation process is performed on each configuration item to be processed to obtain a set of word chunks to be processed, and the sequence start identifier and the sequence end identifier are added to the word chunks to be processed to obtain the word chunk sequence to be processed. Exemplarily, a language model such as BERT (Bidirectional Encoder Representations from Transformers, bidirectional encoder representations based on the Transformer (transformer network model)) can be used to automatically generate corresponding semantic sequences for the source configuration item, the common key, and the configuration item to be processed. Taking the configuration item name of the source configuration item BMC_NTP_Servers1 as an example, the BERT model will perform word segmentation on the source configuration item to obtain BMC, _, NTP, _, Servers, 1, and combine each word chunk to obtain the word chunk sequence , , where n is the number of word chunks, and the sequence start identifier [CLS] and the sequence end identifier [SEP] are added to the sequence to obtain the final source word chunk sequence w: .

[0029] Exemplarily, this embodiment also provides a calculation process for the semantic quantization values of each token block: Each source token block sequence is processed according to the following method. For the sake of description, the source token block sequence being processed is defined as the current source token block sequence. According to the initial values of each token block in the current source token block sequence, the initial value can be assigned according to a pre-set initial value generation method, such as a random number, or can be set in other ways, such as a default value. The position quantization value of each token block in the current source token block sequence determines the initial semantic quantization value of each token block in the current source token block sequence. The position quantization value is used to represent the position of the token block in the sequence, which is the position representation of the token block in the sequence and can use the position sequence number as the quantization value. Based on the pre-set token block initial value optimization information, such as multiplying by an adjustment coefficient, etc., the initial semantic quantization value of the current source token block sequence is adjusted to obtain the semantic quantization value of each token block in the current source token block sequence. According to the initial values of each token block in the common token block sequence and their corresponding position quantization values, the initial semantic quantization value of each token block in the common token block sequence is determined, and according to the token block initial value optimization information, the initial semantic quantization value of each token block in the common token block sequence is adjusted to obtain the semantic quantization value of each token block in the common token block sequence. Similarly, for each token block sequence to be processed, according to the initial values of each token block in the current token block sequence to be processed and the position quantization value of each token block in the current token block sequence to be processed, the initial semantic quantization value of each token block in the current token block sequence to be processed is determined; based on the token block initial value optimization information, the initial semantic quantization value of the current token block sequence to be processed is adjusted to obtain the semantic quantization value of each token block in the current token block sequence to be processed.

[0030] To make those skilled in the art more clearly understand the implementation manner of this embodiment, this embodiment takes the NTP configuration as an example. The NTP time server configuration item obtained from the source server is BMC_TIMER_IP1, that is, the source configuration item is BMC_TIMER_IP1, and the public keys include SERVER_NTP_IP1, SERVER_NTP_IP2, MODE, NTPENABLE (NTP configuration takes effect). First, perform a tokenization operation on BMC_TIMER_IP1 and the SERVER_NTP_IP1 of the public key using the WordPiece (sub-word tokenization method) algorithm to obtain the corresponding sequences respectively: and , and the initial semantic quantization values of each token block in the above sequence can be calculated according to the relational expression where represents the initial semantic quantization value of the i-th token block in the sequence w, is the initial value of the token block, Is a position embedding word, that is, a position quantization value, which represents the position of the token in the sequence. For example, when using a language model to calculate the semantic quantization value of a token, the initial value optimization information of the token is optimized according to the self-attention calculation result. For any token in the above sequence, based on the initial semantic quantization value and the weight parameter matrix of the current token, calculate the query vector, key vector, and value vector of the current token; determine the attention score of the current token according to the key vector and the query vector, and calculate the attention dispersion distribution information of the current token according to the attention score; determine the semantic quantization value of the current token according to the attention dispersion distribution information of the current token and the value vectors of other tokens belonging to the same configuration item.

[0031] When the semantic quantization values of each token in the source configuration item and the common key are calculated, any similarity calculation relationship can be used to calculate the similarity degree between the two. For example, for each token in BMC_TIMER_IP1, calculate the similarity with the semantic quantization values of each token in SERVER_NTP_IP1, SERVER_NTP_IP2, MODE, and NTPENABLE respectively. The preset similarity condition is set in advance and can be flexibly adjusted according to the actual situation. It is used to measure whether two tokens are similar. The preset similarity condition can be, for example, that the similarity is greater than 95%, or the similarity threshold is set to 0.95. If it is not lower than this similarity threshold, the preset similarity condition is satisfied. For example, the similarity degree can be calculated through Calculating the similarity degree Indicates the similarity value, where i is the i-th token, n is the total number of tokens, x represents the semantic quantization values of each token in the source token sequence, and y represents the semantic quantization values corresponding to each token in the common token sequence. The closer this result is to 1, the more similar the two configuration items are.

[0032] Furthermore, considering that there may be a situation where the source configuration item cannot be matched with the common key, and there may be a situation where no corresponding configuration item can be matched for the configuration item to be processed in the common key. Based on the above embodiments, the present invention can also determine the corresponding relationship between the source configuration item and the common key and the corresponding relationship between the common key and the configuration item to be processed in the case of non-matching through alternative items and third-party detection. The implementation process is as follows: When at least one source target configuration item in the source server cannot match a chunk with the same semantics in the common key, multiple candidate chunks that meet the preset similar semantic conditions with the source configuration item are selected from the common key; each candidate chunk and the source configuration item are filled into the first target position of the user configuration page, and the user configuration page is displayed; when a chunk selection instruction is received, the target chunk selected from each candidate chunk is associated with the source configuration item. When at least one to-be-processed target configuration item in the target server cannot match a chunk with the same semantics in the common key, multiple candidate chunks that meet the preset similar semantic conditions with the to-be-processed target configuration item are selected from the common key; each candidate chunk and the to-be-processed target configuration item are filled into the first target position of the user configuration page, and the user configuration page is displayed; when a chunk selection instruction is received, the target chunk selected from each candidate chunk is associated with the to-be-processed target configuration item.

[0033] For example, the preset similarity condition is that the similarity threshold exceeds 0.95. When no chunk with a similarity threshold exceeding 0.95 is found, it is considered that the source configuration item or the to-be-processed configuration item does not match the common key successfully. The preset similar semantic condition is to select the top m chunks with the highest similarity as candidate chunks. For example, m is taken as 3 or 5. Of course, other selection methods can also be used as the preset similar semantic condition, and the present invention does not limit this. For example, when calculating the similarity between the chunk NTP of SERVER_NTP_IP1 and each chunk in the common key, if the highest similarity does not exceed 0.95, it is considered that NTP does not match the common key successfully. The similarities between NTP and each chunk in the common key are sorted from high to low, and the top three with the highest similarity values are used as candidate chunks. The user manually calibrates through the user configuration page, and the target chunk selected from each candidate chunk generates a chunk selection instruction and is sent on the user configuration page. For example, if the source configuration item is BMC_LDAP_Group and the target chunk is ldap_group, the chunk selection instruction can be expressed as: "BMC_LDAP_Group is equivalent to ldap_group."

[0034] As can be seen from the above, this embodiment can support the introduction of an artificial calibration feedback interface, support the user to manually calibrate the unrecognized configuration items through the visual interface, and improve the parameter configuration efficiency and accuracy.

[0035] Furthermore, in order to improve the accuracy of server parameter configuration, based on the above embodiment, the present invention also provides a multiple verification process, which may include the following contents: Such as Figure 5As shown, obtain the current configuration parameter content corresponding to each to-be-processed configuration item of the target server; compare the current configuration parameter content corresponding to each to-be-processed configuration item with the corresponding content in the parameter configuration standard template; when there is target configuration parameter content for which the comparison fails and needs to be confirmed, fill the target configuration parameter content and the corresponding target to-be-processed configuration item into the second target position of the user configuration page, and display the user configuration page; when a configuration information adjustment instruction is received, use the new configuration parameter content to update the target configuration parameter content corresponding to the target server.

[0036] When the parameter configuration of the target server is completed, all configuration items of the target server can be obtained through the Redfish interface and compared with the set parameter configuration standard template. The parameter configuration standard template sets the standard parameter format and the allowable value range for each configuration item. Value verification is completed by verifying the parameter format, such as regular matching of IP addresses, and the value range, such as the number of days for log retention from 0 to 365. Based on the parameter configuration standard template, determine whether the setting is successful. If the setting fails, the failure prompt message and the corresponding content can be displayed to the user in a visual manner, and the user manually adjusts the configuration item. For example, if there is an incorrect mapping of the to-be-processed configuration item to the source configuration item, then manually adjust and reconfigure. Further, an exponential backoff retry mechanism can be enabled for the configuration items that fail the verification. That is, after the first failure, wait for 5 seconds and then retry. If it still fails, wait for 10 seconds the second time, and retry at most 3 times. At the same time, automatically switch to the alternate mapping path, such as switching from "ntp_servers" → "Time_Servers" to "time_servers" → "NTP_Sources", effectively improving the recovery rate of single-point configuration failure.

[0037] Further, as Figure 5 shown, it is also possible to verify whether the server configuration is successful from the business logic, which may include the following content: when the current configuration parameter content corresponding to each to-be-processed configuration item matches the parameter configuration standard template, determine the business function of each to-be-processed configuration item in the server; verify the function of each to-be-processed configuration item: obtain the first function parameter value when setting the current configuration parameter content corresponding to the first to-be-processed configuration item of the target server; obtain the source function parameter value of the source server at the current moment when it performs the same function as the first to-be-processed configuration item; obtain the second function parameter value of the target server at the current moment when it performs the same function as the first to-be-processed configuration item; when the first function parameter value, the second function parameter value, and the source function parameter value meet the preset verification conditions, the first to-be-processed configuration item is successfully configured.

[0038] In this embodiment, when the parameter configuration of the target server is completed, whether value verification is performed on the configuration parameters according to the standard template or not, business functional verification can be carried out. When performing function verification, it can be distinguished according to the functions of the configuration items. For example, verify whether the NTP configuration is successfully configured as required to achieve the business function of time synchronization, and whether the alarm configuration is configured as required to achieve the business function of alarming. Taking NTP_SERVER1 as an example, NTP_SERVER1 is the IP (Internet Protocol Address) of the time synchronization server. Obtain the time on the set IP of 192.168.1.1, obtain the current time of the source server again, and finally obtain the current time of the target server. Compare the time differences among the three. If the preset verification conditions are met, such as the NTP time synchronization error < 5 seconds, that is, the first function parameter value, the second function parameter value, and the source function parameter value do not exceed 5 seconds or do not exceed 10s, then the configuration item is successfully configured.

[0039] As can be seen from the above, in this embodiment, value verification is used to verify the parameter format and value range, and function verification is used to verify the configuration validity from the business logic, which can achieve a configuration error interception rate of 99.2% and effectively improve the parameter configuration accuracy of the target server.

[0040] Based on the above embodiment, the present invention also provides another implementation process of the double-layer mapping relationship, which may include the following contents: Train a semantic quantization model. For ease of description, the word chunks in the training process are defined as sample word chunks, and the semantic quantization values corresponding to the sample word chunks are defined as semantic quantization sample values. The model training process is as follows: First, obtain multiple seed word chunks and generate initial numerical vectors for each seed word chunk; obtain a text sample set, and there are labels indicating whether the text samples in the text sample set contain the same semantic word chunks; perform word segmentation on each text sample in the text sample set, and generate initial semantic quantization sample values for the sample word chunks of each text sample according to the position information of each sample word chunk in the corresponding text sample and the initial numerical vector; adjust the corresponding initial semantic quantization sample values according to the self-attention calculation results of the initial semantic quantization sample values of each sample word chunk to obtain semantic quantization sample values for updating the initial numerical vectors of each seed word chunk; iteratively train the semantic quantization model by continuously approaching the semantic quantization sample values of sample word chunks with the same semantics until the preset model iteration update end condition is met, such as the model converges, or the total number of iterations is reached, or the model accuracy is greater than the preset accuracy threshold, to obtain a trained semantic quantization model. Input each word chunk of each source configuration item, common key, and configuration item to be processed into the trained semantic quantization model in the above process, perform corresponding processing on the input according to the semantic quantization model, and output the corresponding semantic quantization value. When the semantic quantization values of each word chunk are obtained, determine the same semantic word chunks by comparing the semantic quantization values of each word chunk.

[0041] Among them, the seed word chunks are common word chunks in different configuration scenarios of the server. Each word chunk can be randomly assigned a 768-dimensional random number vector as the initial numerical vector. For example, NTP corresponds to the initial numerical vector [0, 1, -0.3, 0.7,...], and this initial numerical vector is 768 random numbers, ranging from -1 to 1. Then, adjust the corresponding 768-dimensional vector through large-scale text training. The text sample set can include billions of sentences, and the 768-dimensional vectors of the seed word chunks are optimized through these sentences. For example, sentence 1 is: NTP server is a general term for setting the server time and time zone, and sentence 2 is: The TIMER server function includes setting the server time and time zone. At this time, NTP server and time server are used as two subjects, and when performing semantic analysis, the semantic similarity is very high, and both NTP and TIMER are in the seed word chunk library, so the 768-dimensional vectors of the two need to gradually approach. Exemplarily, a seed word chunk library can be set in advance. The seed word chunk library can be constructed, for example, through domain ontology modeling technology, and contains at least more than 1200 configuration items, covering data required for at least 23 types of server configuration scenarios such as NTP, logs, and alarms. Further, the system can record the seed scenario library of each user site, record the accuracy rate of each seed scenario library working in the user site, extract the seed scenario libraries with high accuracy rates into the seed word chunk library, and when a new user uses it, use the seed word chunk library with high accuracy as the initialized seed word chunk library without repeated training.

[0042] Exemplarily, after obtaining the text samples, each text sample can be segmented, and a sequence start identifier and a sequence end identifier can be added to each group of text sample chunks to obtain each text chunk sequence; for each text chunk sequence, according to the segment identifier quantization value of the text sample to which the current text chunk sequence belongs, the position quantization value of each sample chunk in the current text chunk sequence corresponding to the current text chunk sequence, and the current initial numerical vector of the seed chunk corresponding to each sample chunk of the current text chunk sequence, the initial semantic quantization sample value of each sample chunk of the current text chunk sequence is determined. For example, the chunk sequence of sentence 1 is , , where n is the number of chunks, is the sequence identifier [CLS] of sentence 1, is the chunk NTP of sentence 1 whose vector needs to be optimized. The chunk sequence of sentence 2 is , , where n is the number of chunks, is the sequence identifier [CLS] of sentence 2, is the chunk TIMER in sentence 2 whose vector needs to be optimized. The initial value of the vector of the j-th chunk in the w-th sequence can be determined according to , where is the initial vector value of the seed chunk, is the segment identifier quantization value, that is, the segment embedding word. This value is a random vector. Each sentence will be randomly initialized with a 768-dimensional vector, representing which sentence it belongs to and used to represent the sentence vector. is the position embedding word, that is, the position quantization value, indicating the position of the chunk in the sentence, that is, when the sentence is segmented, the position of the word in the sentence. For example, the position of NTP in sentence 1 is 1, and the position quantization value is a 768-dimensional vector with a value of 1.

[0043] Exemplarily, the present invention also provides an exemplary network model structure of a semantic quantization model, which may include an input layer, a token processing model layer, an attention calculation layer, a multi-layer feedforward network layer, and an output layer; various sub-tokens and a text sample set are input through the input layer; there are labels indicating whether the text samples in the text sample set contain the same semantic tokens; the token processing model layer is used to generate initial numerical vectors for various sub-tokens, perform word segmentation on each text sample in the text sample set, and generate initial semantic quantization sample values of the sample tokens of each text sample according to the position information and initial numerical vectors of each sample token in the corresponding text sample; the attention calculation layer is used to perform self-attention calculation on each sample token, and adjust the corresponding initial semantic quantization sample value according to the self-attention calculation result of the initial semantic quantization sample value of each sample token, so as to obtain semantic quantization sample values for updating the initial numerical vectors of various sub-tokens; the multi-layer feedforward network layer is used to approximate the semantic quantization sample values of the sample tokens with the same semantics until the preset model iteration update end condition is met.

[0044] Among them, the token processing model layer can adopt, for example, the BERT model, and the attention calculation layer can adopt a multi-layer Transformer network structure. The BERT model is used to perform word segmentation on the configuration item name (such as BMC_NTP_Servers1) to generate initial semantic quantization sample values, and then input them into the multi-layer Transformer network structure. Through the multi-layer Transformer network structure, they are adjusted to generate multi-dimensional vectors containing semantic and order information, and the dimension is the same as the dimension of the initial vector value, such as a 768-dimensional vector. That is, the configuration item is first encoded by BERT. , where, represents the output of BERT, , , starts as the token after word segmentation. In the BERT model, the output of the last layer contains rich semantic information of the entire configuration item column name. Among them, the vector corresponding to the [CLS] token , C represents the vector value of [CLS], which is used as the semantic vector of the entire column name for subsequent classification, similarity calculation and other tasks. The result of the first encoding is used as the encoding input parameter for the second encoding, and the formula is , represents the number of layers, and finally the semantic vector of each configuration item is obtained, that is, the semantic quantization sample value. Through layer-by-layer encoding, the output of each layer is the intermediate feature representation of the input sequence encoded by this layer. As the number of layers of the multi-layer Transformer network structure increases, gradually contains richer and more abstract semantic information. For example, in the lower layer of Among them, the local features and shallow semantics of the chunks may be captured more, while in the higher-level among them, the global semantics of the entire sequence and complex dependencies can be captured. The output of each layer will be used as the input of the next layer , and through layer-by-layer transmission and processing, the final result is obtained.

[0045] Exemplarily, the secondary encoding process of the attention calculation layer is as follows: for each sample chunk, based on the initial semantic quantization sample value and weight parameter matrix of the current sample chunk, calculate the query vector, key vector, and value vector of the current sample chunk; determine the attention score of the current sample chunk according to the key vector and query vector, and calculate the attention dispersion distribution information of the current sample chunk according to the attention score; determine the semantic quantization sample value of the current sample chunk according to the attention dispersion distribution information of the current sample chunk and the value vectors of other sample chunks belonging to the same text sample.

[0046] For example, the initial semantic quantization sample value of the sample chunk is , and the weight parameter matrix is a learnable parameter, which are respectively represented as , In the BERT model, conventional initial values will be given, and these parameters can be continuously learned and adjusted during the model training process. The query vector, key vector, and value vector are Q, K, and V, and can be calculated according to the relational expressions , , respectively to calculate the query vector, key vector, and value vector. The attention score AS can be calculated based on the relational expression , is the dimension of the query vector and key vector, and the initial learning matrix is a 64-dimensional learning matrix, the initial value can be set to 8, the softmax function can be used, that is, AD = softmax(AS), to calculate the attention dispersion distribution AD, and finally, the output vector after secondary processing, that is, the semantic quantization sample value can be obtained through , is the value vector calculated for other word segments in the sentence. At this time, the semantic quantization sample values of word chunks with the same semantics, such as the semantic quantization sample value of the NTP configuration item in sentence 1 in the above example , the semantic quantization sample value of the TIMER configuration item in sentence 2 in the above example , begin to have a tendency to be close. Finally, the vector after secondary processing is adjusted by a feed-forward network, that is, the output vector O(w) is input into the feed-forward network layer. The initial value of the feed-forward network layer can be 12 layers and is dynamically adjusted according to the sample data and the model training situation. The feed-forward network layer is processed through the relational expression , and is a learnable weight matrix, both are bias vectors, which are parameters to be learned during model training. ReLU (Linear rectification function) is used as the activation function for the feedforward network layer, FN is the output of the feedforward network layer, and this process is continuously repeated for multiple feedforward network layers until the two vectors continuously approach each other.

[0047] Furthermore, to improve the performance of the semantic quantization model, the semantic quantization model can also be adjusted during use, which may include the following: Take the mapping relationship between the configuration items of the successfully configured target server and the source server as a new training sample; when a seed update instruction is triggered, if there are new seed chunks in the new training sample that do not exist in the seed chunk library, add the new seed chunks to the seed chunk library and use the semantic quantization sample values of the new seed chunks as the initial numerical vectors; if the semantic quantization sample values of the chunks in the new training sample are different from the initial numerical vectors of the same seed chunks in the seed chunk library, use the semantic quantization sample values of the chunks in the new training sample to update the initial numerical vectors of the corresponding chunks.

[0048] In this embodiment, after all the configurations of the source server are copied to the target server, the source configuration items, common keys, and each chunk of the to-be-processed configuration items are compared with the seed chunks to determine whether there are new seed chunks that are not in the seed chunk library used during model training. If so, the seed library is updated. Further, the situation of different initial numerical vectors for the same seed chunk can also be updated. For the corresponding relationships established between the source configuration items and the common keys, between the common keys and the to-be-processed configuration items, and between the source configuration items and the to-be-processed configuration items during this mapping process, a semantic sentence can be added for the already paired configuration items. For example, SERVER_NTP_IP1 is equivalent to BMC_TIMER_IP1, and the generated statement is put back into the text sample set and trained with the new samples to improve the accuracy of the model. For the accuracy of the data samples, after the value verification and function verification of the target server are completed, the corresponding semantic sentences can be generated for the configuration items with established mapping relationships. Further, to improve the performance of the semantic quantization model, the manual calibration results in the above embodiments or the content manually adjusted, such as "SERVER_NTP_IP1 is equivalent to BMC_TIMER_IP1", "BMC_LDAP_Group is equivalent to ldap_group", can be converted into semantic sentences. After being parsed by the natural language model, the seed chunk library or the text sample set is updated, and the semantic quantization model is retrained regularly or triggered. For example, when the performance of the semantic quantization model decreases, a retraining is triggered, or when the new seeds or new text samples exceed a certain threshold, a training is triggered. Through active learning, the accuracy of the model increases linearly with the usage time. Through the closed loop of "manual intervention - model learning - automatic optimization", the efficiency and accuracy of server parameter configuration are improved.

[0049] Finally, the present invention also provides another implementation method for server configuration. In this embodiment, an end-to-end intelligent semantic entity is pre-trained. The intelligent semantic entity at least includes a data collection tool, a seed chunk library, a semantic quantization model, a semantic recognition module, and a verification module. Among them, the semantic recognition module encapsulates a computer program for performing similarity calculation and selecting the same chunks according to the similarity calculation results. The verification module encapsulates a computer program for performing value verification and function verification. Through this end-to-end intelligent semantic entity, the configuration process of each to-be-processed configuration item of the target server can be directly completed: A11: Input the authentication basic information such as the username and password of the on-site server. Use the data collection tool to collect the configuration items of all servers, generate standard configuration parameter subclasses and source configuration information, and convert all the obtained configuration information into the key - value format.

[0050] A12: Segment each source configuration item information of the source server one by one. Based on the chunk position and the initial value of the chunk, calculate the initial semantic quantization value of each chunk in each source chunk sequence. Perform self-attention calculation on all the segmented words of the configuration item to obtain the semantic quantization value of each data chunk.

[0051] A13: Also segment and perform attention calculation on the public key, and then perform encoder encoding to generate the semantic quantization value of each public key.

[0052] A14: Loop through the semantic quantization values of all the configuration items of the source server, calculate the similarity between each semantic quantization value and the semantic quantization value of the public key, find the item in the public key whose similarity exceeds 0.95 for the source configuration item, and use it as the chunk with the same semantics. That is, the source configuration item is successfully mapped in the public key. For the source configuration item that fails to map, it can be marked and sent to the user in a visual way for manual calibration.

[0053] A15: After the mapping of the source service configuration items is completed, collect the configuration items to be processed of the target server. Segment, perform attention calculation, and encoder encoding on all the configuration items of the target server in the same way to generate the semantic quantization sample values of all the configuration items of the target server. Loop through the semantic quantization sample values of all the configuration items of the target server, calculate the similarity between each semantic quantization sample value and the semantic vector of the public key, find the item in the public key whose similarity exceeds 0.95 for the configuration item to be processed, and use it as the chunk with the same semantics. That is, the configuration item to be processed is successfully mapped in the public key. For the configuration item to be processed that fails to map, it can be marked and sent to the user in a visual way for manual calibration.

[0054] A16: After the double-layer mapping is completed, through the public key, configure the content of the configuration items of the source server as the content of the corresponding configuration items of the target server, map all the content of the source server configuration items one by one to the required configuration content of the target server configuration items, convert this format into an object, and call the Redfish interface to perform corresponding configuration on the configuration items to be processed of the target server.

[0055] A17: After the configuration is completed, perform value verification and functional verification on the configuration results respectively, and feedback the configuration failure situation to the user so that the user can make manual adjustments.

[0056] A18: Record the whole process of the configuration item adjustment. For the finally correct mapping result, feedback it to the intelligent semantic body, and the intelligent semantic body will regenerate the entire seed chunk library.

[0057] As can be seen from the above, in this embodiment, the intelligent semantic body is constructed to drive the server for configuration. The configuration items of the source server are automatically collected through the Redfish protocol. Combining the BERT semantic vector generation and double-layer mapping, the replication time of the single-server configuration is shortened from more than 30 minutes of manual operation to the minute level, and the efficiency is increased by more than 90%. The 768-dimensional semantic vector based on the BERT model captures the deep meaning of the configuration items, solves the problems of the same name with different meanings and different names with the same meaning, and the configuration error rate is reduced from 15% of the traditional solution to less than 5%. Through double-value verification and function verification, it is ensured that the configuration takes effect and meets the business expectations, avoiding the abnormal server functions caused by configuration errors, realizing the automation and intelligence of cross-vendor server configuration, and achieving efficient and accurate server configuration.

[0058] It should be noted that there is no strict order of execution among the steps of the present invention. As long as it conforms to the logical order, these steps can be executed simultaneously or in a certain preset order. Figure 2 and Figure 5 This is just a schematic way and does not mean that it can only be such an execution order.

[0059] Through the description of the above embodiments, those skilled in the art can clearly understand that the method according to the above embodiments can be implemented by means of software plus a necessary general hardware platform. Of course, it can also be implemented by hardware, but in many cases, the former is a better implementation method.

[0060] The present invention also provides a corresponding device for the server parameter configuration method, which further makes the method more practical. Among them, the device can be described from the perspective of functional modules and the perspective of hardware respectively. The server parameter configuration device provided by the present invention will be introduced below. The device is used to implement the server parameter configuration method provided by the present invention. In this embodiment, the server parameter configuration device can include or be divided into one or more program modules. The one or more program modules are stored in the storage medium and executed by one or more processors to complete the server parameter configuration method disclosed in Embodiment 1. The program modules referred to in this embodiment refer to a series of computer program instruction segments that can complete specific functions, and are more suitable for describing the execution process of the server parameter configuration device in the storage medium than the program itself. The following description will specifically introduce the functions of each program module in this embodiment. The server parameter configuration device described below can be mutually corresponded and referred to with the server parameter configuration method described above.

[0061] From the perspective of functional modules, see Figure 6 , Figure 6 which is the structural diagram of the server parameter configuration device provided by this embodiment in a specific implementation manner. The device can include: The public standard information construction module 601 is used to re-define the to-be-configured information of the servers in the data center into multiple standard configuration parameter subclasses, and construct public key-values according to the standard configuration parameter subclasses, where each standard configuration parameter subclass corresponds to at least one configuration item.

[0062] The first-level mapping module 602 is used to obtain each source configuration item of the source server with configured parameters, determine the corresponding relationship between each source configuration item and the public key by identifying the same semantic phrase blocks of the public keys of each source configuration item and the public key-value, and map the configuration content of each source configuration item to the value corresponding to the public key based on the condition that the configuration contents of the configuration items with the same semantics are the same.

[0063] The second-level mapping module 603 determines the corresponding relationship between each to-be-processed configuration item and the public key by identifying the same semantic phrase blocks of each to-be-processed configuration item and the public key in the target server, and generates the corresponding configuration content for each to-be-processed configuration item based on the mapped public key-value to complete the parameter configuration of the target server.

[0064] Exemplarily, in some implementation manners of this embodiment, the above first-level mapping module 602 can also be used to: respectively extract each word block of each source configuration item and the public key, and generate the corresponding source word block sequence and public word block sequence; calculate the semantic quantization values of each word block in each source word block sequence and each public word block sequence respectively based on the word block position and the initial value of the word block; calculate the similarity degree between the semantic quantization values of each word block in each source word block sequence and the semantic quantization values of each word block in the public word block sequence, and use two word blocks that meet the preset similarity condition as the same semantic phrase blocks.

[0065] As an exemplary implementation manner of the above embodiment, the above first-level mapping module 602 can further be used to: perform word segmentation processing on each source configuration item to obtain a group of source configuration word blocks corresponding to each source configuration item, add a sequence start identifier and a sequence end identifier to each group of source configuration word blocks to obtain each source word block sequence; perform word segmentation processing on the public key to obtain a group of public word blocks, and add a sequence start identifier and a sequence end identifier to the public word blocks to obtain the public word block sequence.

[0066] As another exemplary implementation of the above embodiment, the first-level mapping module 602 may further be configured to: for each source word block sequence, determine the initial semantic quantization values of the word blocks in the current source word block sequence according to the initial values of the word blocks in the current source word block sequence and the position quantization values of the word blocks in the current source word block sequence; based on the word block initial value optimization information, adjust the initial semantic quantization values of the current source word block sequence to obtain the semantic quantization values of the word blocks in the current source word block sequence; according to the initial values of the word blocks in the common word block sequence and their corresponding position quantization values, determine the initial semantic quantization values of the word blocks in the common word block sequence, and adjust the initial semantic quantization values of the word blocks in the common word block sequence according to the word block initial value optimization information to obtain the semantic quantization values of the word blocks in the common word block sequence.

[0067] Exemplarily, in some other embodiments of this embodiment, the second-level mapping module 603 may further be configured to: when at least one target configuration item to be processed in the target server cannot match a word block with the same semantics in the common key, select multiple candidate word blocks that meet the preset similar semantic conditions with the target configuration item to be processed from the common key; fill each candidate word block and the target configuration item to be processed into the first target position of the user configuration page, and display the user configuration page; when receiving a word block selection instruction, establish a correspondence relationship between the selected target word block from each candidate word block and the target configuration item to be processed.

[0068] Exemplarily, in some other embodiments of this embodiment, the apparatus further includes a verification module, configured to obtain the current configuration parameter content corresponding to each target configuration item to be processed of the target server; compare the current configuration parameter content corresponding to each target configuration item to be processed with the corresponding content of the parameter configuration standard template; when there is target configuration parameter content to be confirmed with a comparison failure, fill the target configuration parameter content and the corresponding target configuration item to be processed into the second target position of the user configuration page, and display the user configuration page; when receiving a configuration information adjustment instruction, update the target configuration parameter content corresponding to the target server with the new configuration parameter content.

[0069] As an exemplary implementation of the above embodiments, the above verification module may further be configured to: when the current configuration parameter content corresponding to each to-be-processed configuration item matches the parameter configuration standard template, determine the business functions of each to-be-processed configuration item in the server; perform function verification on each to-be-processed configuration item: obtain the first function parameter value when setting the current configuration parameter content corresponding to the first to-be-processed configuration item on the target server; obtain the source function parameter value of the source server when the first to-be-processed configuration item performs the same function at the current moment; obtain the second function parameter value of the target server when the first to-be-processed configuration item performs the same function at the current moment; when the first function parameter value, the second function parameter value, and the source function parameter value meet the preset verification conditions, the first to-be-processed configuration item is successfully configured.

[0070] Exemplarily, in some other implementation manners of this embodiment, the above first-level mapping module 602 may further be configured to: input each word block of each source configuration item and the common key into the trained semantic quantization model respectively, obtain the semantic quantization value of each word block according to the output of the semantic quantization model, and determine the same semantic word blocks by comparing the semantic quantization values of each word block; wherein, the training process of the semantic quantization model includes: pre-obtaining a plurality of seed word blocks and generating initial numerical vectors for each seed word block; obtaining a text sample set, and each text sample in the text sample set has a label indicating whether it contains the same semantic word blocks; performing word segmentation on each text sample in the text sample set, and generating the initial semantic quantization sample value of the sample word blocks of each text sample according to the position information of each sample word block in the corresponding text sample and the initial numerical vector; adjusting the corresponding initial semantic quantization sample value according to the self-attention calculation result of the initial semantic quantization sample value of each sample word block to obtain the semantic quantization sample value for updating the initial numerical vector of each seed word block; iteratively training the semantic quantization model by continuously approaching the semantic quantization sample values of the sample word blocks with the same semantics until the preset model iteration update end condition is met, and obtaining the trained semantic quantization model.

[0071] As an exemplary implementation of the above embodiments, the semantic quantization model includes an input layer, a token processing model layer, an attention calculation layer, a multi-layer feedforward network layer, and an output layer; various sub-tokens and a text sample set are input through the input layer; there are labels indicating whether the text samples in the text sample set contain the same semantic tokens; the token processing model layer is used to generate initial numerical vectors for various sub-tokens, perform word segmentation on each text sample in the text sample set, and generate initial semantic quantization sample values of the sample tokens in each text sample according to the position information and initial numerical vectors of each sample token in the corresponding text sample; the attention calculation layer is used to perform self-attention calculation on each sample token, and adjust the corresponding initial semantic quantization sample value according to the self-attention calculation result of the initial semantic quantization sample value of each sample token, so as to obtain semantic quantization sample values for updating the initial numerical vectors of various sub-tokens; the multi-layer feedforward network layer is used to approximate the semantic quantization sample values of the sample tokens with the same semantics until the preset model iteration update end condition is met.

[0072] As another exemplary implementation of the above embodiments, the first-level mapping module 602 may further be used for: performing word segmentation on each text sample, and adding a sequence start identifier and a sequence end identifier to each group of text sample tokens to obtain each text token sequence; for each text token sequence, determining the initial semantic quantization sample value of each sample token in the current text token sequence according to the segment identifier quantization value of the text sample to which the current text token sequence belongs, the position quantization value of each sample token in the current text token sequence corresponding to the current text token sequence, and the current initial numerical vector of the seed token corresponding to each sample token in the current text token sequence.

[0073] As another exemplary implementation of the above embodiments, the first-level mapping module 602 may further be used for: for each sample token, calculating the query vector, key vector, and value vector of the current sample token based on the initial semantic quantization sample value and weight parameter matrix of the current sample token; determining the attention score of the current sample token according to the key vector and query vector, and calculating the attention dispersion distribution information of the current sample token according to the attention score; determining the semantic quantization sample value of the current sample token according to the attention dispersion distribution information of the current sample token and the value vectors of other sample tokens belonging to the same text sample.

[0074] As another exemplary implementation of the above embodiments, the first-level mapping module 602 may further be configured to: use the mapping relationship between the configuration items of the target server with successful configuration and the source server as a new training sample; when a seed update instruction is triggered, if there are new seed chunks in the new training sample that do not exist in the seed chunk library, add the new seed chunks to the seed chunk library, and use the semantic quantization sample value of the new seed chunks as the initial numerical vector; if the semantic quantization sample value of the chunks in the new training sample is different from the initial numerical vector of the same seed chunks in the seed chunk library, update the initial numerical vector of the corresponding chunks with the semantic quantization sample value of the chunks in the new training sample.

[0075] The server parameter configuration device mentioned above is described from the perspective of functional modules. Further, the present invention also provides an electronic device, which is described from the perspective of hardware. Figure 7 The following is a schematic structural diagram of the electronic device provided by the embodiment of the present invention in an implementation manner. The electronic device includes a memory 701 and a processor 702. A computer program is stored in the memory 701, and the processor 702 is configured to run the computer program to execute the steps in any of the above-mentioned embodiments of the server parameter configuration method.

[0076] The embodiment of the present application also provides a computer-readable storage medium, in which a computer program is stored. The computer program is configured to execute the steps in any of the above-mentioned embodiments of the server parameter configuration method when running.

[0077] [[ID=**11**]]In an exemplary embodiment, the above computer-readable storage medium may include, but is not limited to: USB flash drive, read-only memory (ROM for short), random access memory (RAM for short), mobile hard disk, magnetic disk, or optical disc, etc., various media that can store computer programs.

[0078] The embodiment of the present application also provides a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, it implements the steps in any of the above-mentioned embodiments of the server parameter configuration method.

[0079] The embodiment of the present application also provides another computer program product, including a non-volatile computer-readable storage medium. The non-volatile computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it implements the steps in any of the above-mentioned embodiments of the server parameter configuration method.

[0080] The above has introduced in detail a server parameter configuration method, an electronic device, a computer-readable storage medium, and a computer program product provided by the present invention. Each embodiment in this specification is described in a progressive manner. The key point of each embodiment is the difference from other embodiments. For the same or similar parts between the embodiments, reference can be made to each other. Whether the units and algorithm steps of each example described in the disclosed embodiments are executed in the form of electronic hardware or computer software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, and such implementation should not be considered to exceed the scope of the present invention. Without departing from the principle of the present invention, several improvements and modifications can also be made to the present invention, and these improvements and modifications also fall within the protection scope of the present invention.

Claims

1. A server parameter configuration method, characterized in that, Including: Redefine the configuration information to be configured of the servers in the data center into multiple standard configuration parameter subclasses, and construct a common key-value based on each standard configuration parameter subclass, where each standard configuration parameter subclass corresponds to at least one configuration item; Obtain each source configuration item of the source server with configured parameters, determine the corresponding relationship between each source configuration item and the common key of the common key-value by identifying the same semantic chunks of each source configuration item and the common key of the common key-value, and map the configuration content of each source configuration item to the value corresponding to the common key based on the condition that the configuration contents of the configuration items with the same semantics are the same; Determine the corresponding relationship between each to-be-processed configuration item and the common key by identifying the same semantic chunks of each to-be-processed configuration item and the common key in the target server, and generate the corresponding configuration content for each to-be-processed configuration item based on the mapped common key-value to complete the parameter configuration of the target server.

2. The server parameter configuration method according to claim 1, characterized in that Determine the corresponding relationship between each source configuration item and the common key by identifying the same semantic chunks of each source configuration item and the common key of the common key-value, including: Extract each chunk of each source configuration item and the common key respectively, and generate the corresponding source chunk sequence and common chunk sequence; Based on the chunk position and the initial value of the chunk, calculate the semantic quantization value of each chunk in each source chunk sequence and the common chunk sequence respectively; Calculate the similarity degree between the semantic quantization value of each chunk in each source chunk sequence and the semantic quantization value of each chunk in the common chunk sequence respectively, and use two chunks that meet the preset similarity condition as the same semantic chunks.

3. The server parameter configuration method according to claim 2, wherein Extract each chunk of each source configuration item and the common key respectively, and generate the corresponding source chunk sequence and common chunk sequence, including: Perform word segmentation on each source configuration item to obtain a set of source configuration chunks corresponding to each source configuration item, and add a sequence start identifier and a sequence end identifier to each set of source configuration chunks to obtain each source chunk sequence; Perform word segmentation on the common key to obtain a set of common chunks, and add a sequence start identifier and a sequence end identifier to the common chunks to obtain the common chunk sequence.

4. The server parameter configuration method according to claim 2, wherein Based on the chunk position and the initial value of the chunk, calculate the semantic quantization value of each chunk in each source chunk sequence and the common chunk sequence respectively, including: For each source chunk sequence, determine the initial semantic quantization value of each chunk in the current source chunk sequence according to the initial value of each chunk in the current source chunk sequence and the position quantization value of each chunk in the current source chunk sequence; based on the chunk initial value optimization information, adjust the initial semantic quantization value of the current source chunk sequence to obtain the semantic quantization value of each chunk in the current source chunk sequence; Determine the initial semantic quantization value of each chunk in the common chunk sequence according to the initial value of each chunk in the common chunk sequence and its corresponding position quantization value, and adjust the initial semantic quantization value of each chunk in the common chunk sequence according to the chunk initial value optimization information to obtain the semantic quantization value of each chunk in the common chunk sequence.

5. The server parameter configuration method according to claim 1, wherein Determining the corresponding relationship between each to-be-processed configuration item and the common key by identifying the same semantic chunks of each to-be-processed configuration item and the common key in the target server, including: When at least one to-be-processed target configuration item in the target server cannot match a chunk with the same semantics in the common key, select multiple candidate chunks that meet the preset similar semantic conditions with the to-be-processed target configuration item from the common key; Fill each candidate chunk and the to-be-processed target configuration item into the first target position of the user configuration page, and display the user configuration page; When receiving a chunk selection instruction, establish a corresponding relationship between the selected target chunk from each candidate chunk and the to-be-processed target configuration item.

6. The server parameter configuration method according to claim 1, characterized in that After generating corresponding configuration content for each to-be-processed configuration item based on the mapped common key-value, further including: Obtain the current configuration parameter content corresponding to each to-be-processed configuration item of the target server; Compare the current configuration parameter content corresponding to each to-be-processed configuration item with the corresponding content of the parameter configuration standard template; When there is target configuration parameter content to be confirmed with a comparison failure, fill the target configuration parameter content and the corresponding target to-be-processed configuration item into the second target position of the user configuration page, and display the user configuration page; When receiving a configuration information adjustment instruction, update the target configuration parameter content corresponding to the target server with the new configuration parameter content.

7. The server parameter configuration method according to claim 6, characterized in that, After comparing the current configuration parameter content corresponding to each to-be-processed configuration item with the corresponding content of the parameter configuration standard template, further including: When the current configuration parameter content corresponding to each to-be-processed configuration item matches the parameter configuration standard template successfully, determine the business functions of each to-be-processed configuration item in the server; Function verification for each to-be-processed configuration item: Obtain the first function parameter value when setting the current configuration parameter content corresponding to the first to-be-processed configuration item of the target server; Obtain the source function parameter value of the source server at the current moment when it has the same function as the first to-be-processed configuration item; Obtain the second function parameter value of the target server at the current moment when it has the same function as the first to-be-processed configuration item; When the first function parameter value, the second function parameter value, and the source function parameter value meet the preset verification conditions, the first to-be-processed configuration item is successfully configured.

8. The server parameter configuration method according to any one of claims 1 to 7, characterized in that Determining the corresponding relationship between each source configuration item and the common key by identifying the same semantic chunks of each source configuration item and the common key of the common key-value, including: Input each source configuration item and each chunk of the common key into the trained semantic quantization model respectively, obtain the semantic quantization value of each chunk according to the output of the semantic quantization model, and determine the same semantic chunks by comparing the semantic quantization values of each chunk; Among them, the training process of the semantic quantization model includes: Pre-obtain multiple seed chunks, and generate initial numerical vectors for each seed chunk; Obtain a text sample set, and each text sample in the text sample set has a label indicating whether it contains the same semantic chunks; Tokenize each text sample in the text sample set, and generate the initial semantic quantization sample values of the sample word chunks of each text sample according to the position information and the initial numerical vectors of the sample word chunks in the corresponding text samples; Adjust the corresponding initial semantic quantization sample values according to the self-attention calculation results of the initial semantic quantization sample values of the sample word chunks, and obtain the semantic quantization sample values for updating the initial numerical vectors of various sub-word chunks; Iteratively train the semantic quantization model by continuously approaching the semantic quantization sample values of the sample word chunks with the same semantics until the preset model iteration update end condition is satisfied, and obtain the trained semantic quantization model.

9. The server parameter configuration method according to claim 8, characterized in that The semantic quantization model includes an input layer, a word chunk processing model layer, an attention calculation layer, a multi-layer feed-forward network layer, and an output layer; Input various sub-word chunks and the text sample set through the input layer; there are labels indicating whether the text samples in the text sample set contain the same semantic word chunks; Use the word chunk processing model layer to generate initial numerical vectors for various sub-word chunks, tokenize each text sample in the text sample set, and generate the initial semantic quantization sample values of the sample word chunks of each text sample according to the position information and the initial numerical vectors of the sample word chunks in the corresponding text samples; Use the attention calculation layer to perform self-attention calculation on the sample word chunks, and adjust the corresponding initial semantic quantization sample values according to the self-attention calculation results of the initial semantic quantization sample values of the sample word chunks, and obtain the semantic quantization sample values for updating the initial numerical vectors of various sub-word chunks; Use each layer of the feed-forward network layer to approximate the semantic quantization sample values of the sample word chunks with the same semantics until the preset model iteration update end condition is satisfied.

10. The server parameter configuration method according to claim 9, wherein Generating the initial semantic quantization sample values of the sample word chunks of each text sample according to the position information and the initial numerical vectors of the sample word chunks in the corresponding text samples includes: Tokenize each text sample, and add a sequence start identifier and a sequence end identifier to each group of sample word chunks to obtain each text word chunk sequence; For each text word chunk sequence, determine the initial semantic quantization sample values of the sample word chunks in the current text word chunk sequence according to the segment identifier quantization value of the text sample to which the current text word chunk sequence belongs, the position quantization values of the sample word chunks in the current text word chunk sequence corresponding to the current text word chunk sequence, and the current initial numerical vectors of the seed word chunks corresponding to the sample word chunks in the current text word chunk sequence.

11. The server parameter configuration method according to claim 9, characterized in that, Adjusting the corresponding initial semantic quantization sample values according to the self-attention calculation results of the initial semantic quantization sample values of the sample word chunks includes: For each sample word chunk, calculate the query vector, key vector, and value vector of the current sample word chunk based on the initial semantic quantization sample value of the current sample word chunk and the weight parameter matrix; Determine the attention score of the current sample word chunk according to the key vector and the query vector, and calculate the attention dispersion distribution information of the current sample word chunk according to the attention score; Determine the semantic quantization sample value of the current sample token according to the attention dispersion distribution information of the current sample token and the value vectors of other sample tokens belonging to the same text sample.

12. The server parameter configuration method according to claim 9, wherein After generating corresponding configuration contents for each configuration item to be processed based on the mapped common key-value, it further includes: Taking the mapping relationship between the configuration items of the target server with successful configuration and the source server as a new training sample; When a seed update instruction is triggered, if there are new seed tokens in the new training sample that do not exist in the seed token library, add the new seed tokens to the seed token library and use the semantic quantization sample value of the new seed tokens as the initial numerical vector; If the semantic quantization sample value of the token in the new training sample is different from the initial numerical vector of the same seed token in the seed token library, use the semantic quantization sample value of the token in the new training sample to update the initial numerical vector of the corresponding token.

13. An electronic device, characterized in that, It includes: A memory for storing computer programs; A processor for implementing the steps of the server parameter configuration method according to any one of claims 1 to 12 when executing the computer program.

14. A computer-readable storage medium, characterized in that, A computer program is stored on the computer-readable storage medium, and when the computer program is executed by the processor, it implements the steps of the server parameter configuration method according to any one of claims 1 to 12.

15. A computer program product, comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, it implements the steps of the server parameter configuration method according to any one of claims 1 to 12.

Images