An efficient cloud network configuration information processing method

Abstract

The application discloses a kind of efficient cloud network configuration information processing method, guarantee the history information in any time period can be accurately restored.The application saves all history information in the form of string in memory by hash and exclusive or etc..In the string snapshot of two time points separated by limited time, it can be judged whether a piece of information belongs to the history information obtained in the time period.In the configuration verification in cloud network, the method can exclude most of the errors caused by time sequence with very low additional cost, and ensure the service quality of network.Compared with existing methods, the method can reflect the history information in any limited length time in real time, thereby reducing the residence time of network failure to half of the original, and the additional cost can be ignored.

Description

Technical Field

[0001] This invention relates to the field of cloud network operation and maintenance management, and in particular to a configuration information processing method for deployment in cloud networks that is highly available, low-overhead, and highly accurate. Background Technology

[0002] With the rapid development of cloud computing technology, more and more enterprises and organizations are migrating their businesses and applications to the cloud to gain greater flexibility, reliability, and scalability. The core of cloud computing is the cloud network, which provides a highly configurable and programmable platform that enables elastic resource utilization and visualized management. In the cloud network, network configurations issued by the controller define how network resources operate, how they communicate with each other, and how security management is performed. Some common network configurations include IP address allocation, virtual network settings, security policies, traffic routing, and load balancing. Good network configuration can improve network availability, performance, and security, thereby ensuring stable business operations.

[0003] However, complex and dynamic large-scale networks are difficult to monitor in real time, and cloud network providers cannot guarantee that the configurations they distribute will always be consistent with expectations. When configuration errors occur, they can have a devastating impact on the service quality of specific users and even the entire network, causing serious network failures such as traffic black holes. Therefore, in traditional operation and maintenance management, cloud network providers need to periodically verify whether the configurations applied on network elements are the same as the configurations expected by the control plane. Note that this process includes not only the verification received by the controller and network elements, but also the verification between the configuration received by the network elements from the control plane and the actual configuration applied to the network elements. However, because the configuration distribution process is strictly ordered, all configurations must be distributed from the control plane to the data plane, and then applied to the backend programs by the network elements received by the data plane. This process means that configuration consistency verification is not performed completely simultaneously. The inconsistencies obtained during the verification process include not only incorrectly distributed configurations, but also parts that did not take effect in the previous step. This part is defined as an error caused by timing inconsistencies. For example, at 18:00:00, the controller compresses the historical configuration information S that needs to be verified and initiates a configuration consistency verification. This information is received by the corresponding network element at 18:00:05, initiating the verification process. During these 5 seconds, the controller continues to continuously send configurations to the network in real time. Information added during these 5 seconds that is not in S is considered inconsistent and is filtered out. Inconsistency at this point is not equivalent to error, and handling these inconsistencies caused by timing issues places a significant burden on the network. When verification is performed internally within the network element, the time interval between these inconsistencies is even longer, typically exceeding 20 seconds, because it requires waiting for the system to extract the configuration from the application's backend program.

[0004] Currently, there are two main methods for handling inconsistencies caused by timing issues. One method is to store all historical information within the relevant time period in local storage. Inconsistencies are then filtered out; any inconsistencies found in the local historical information after the reconciliation initiation point are considered to be due to timing issues and are discarded. This method requires additional storage space. For example, with an average configuration length of 500 bytes and 5000 configurations sent per second, storing historical information per minute requires approximately 143MB of storage space. This size increases further with the average configuration length and network traffic. In fact, considering code implementation and system robustness, the method of recording all historical information does not only record historical information at the time of reconciliation initiation but records all historical information and discards outdated information in stages. For example, retaining 30 minutes of valid information requires 4.2GB of storage on the device. The redundant resources on the controller are sufficient to meet the storage requirements. However, resources on commercial network elements are limited. When handling inconsistency reconciliation within the network element, saving historical information wastes a significant amount of storage resources. In addition, this method requires a precise clock to support the query, which is not necessarily supported on commercial network element devices.

[0005] Another approach is to record all inconsistencies each time without processing them directly. Instead, the intersection of the newly obtained inconsistencies with those from the previous reconciliation is taken, and the intersection is considered the true error. This method requires no additional storage resources, but the latency of the fault doubles, increasing to two reconciliation cycles. For example, if a reconciliation is initiated every 30 minutes, in the worst-case scenario, incorrect configurations could remain on the network for an hour, severely impacting network service quality.

[0006] In summary, the storage of historical information based on cloud network configuration urgently requires a highly available, low-overhead, and highly accurate information storage and restoration method to effectively guarantee the quality of network services. Summary of the Invention

[0007] The purpose of this invention is to address the shortcomings of existing technologies by designing a highly available, low-overhead, and highly accurate cloud network configuration information processing method, which can effectively reduce the overhead for cloud service providers in configuration issues.

[0008] The objective of this invention is achieved through the following technical solution: an efficient cloud network configuration information processing method, which includes the following steps:

[0009] S1, create two strings L1 and L2 in the memory of the network element device, each consisting of m r-bit cells; whenever the network element device receives configuration information from the controller, it stores the configuration information in strings L1 and L2 using a hash function combined with an XOR operation.

[0010] S2, obtain a summary of network configuration history information within a given time period, including: obtaining snapshots S1 and S2 of two strings L1 and L2 at two given time points T1 and T2, and XORing L1 and L2 in snapshots S1 and S2 at different time points respectively to obtain two strings C1 and C2;

[0011] S3, obtain the set of inconsistencies between the configuration issued by the controller and the actual effective configuration of the network element device;

[0012] S4, restore the network configuration history information within a given time period, including: for all elements in the inconsistent set Set, perform an XOR operation on the binary number recorded at the corresponding position in strings C1 and C2 and the binary number calculated by the hash function, and determine whether the corresponding element belongs to the network configuration history information within the given time period based on whether the result is 0; if the element belongs to the network configuration history information, remove the element from strings C1, C2 and set Set.

[0013] Furthermore, the network configuration includes three categories: update, delete, and add. The configuration information is distributed in key-value pair format. The specific content of the three types of configuration is as follows: update: given the content of a key-value pair, the key remains unchanged, but the value is changed; delete: given the content of a key-value pair, delete the corresponding content; add: given the content of a key-value pair, add the corresponding content. When the configuration information is stored as a string, the configuration type is not recorded, only the key-value pair content of the configuration is recorded.

[0014] Furthermore, the two strings L1 and L2 created in the network element device memory are completely identical, consisting of m cells, each cell having a length of r bits. The size of m and r is directly related to the method performance, and the configuration is continuously distributed, so the size of m and r will not change after being set.

[0015] Furthermore, for configuration information in key-value pair format, configuration summary information is recorded through the following process:

[0016] (1) Concatenate the key and value of the configuration information received from the controller to obtain string E. Calculate a hash value for E using hash functions H1 and H2 respectively. Take the remainder of the number of cells m in strings L1 and L2 to obtain the corresponding cell position indices Index1 and Index2.

[0017] (2) Calculate an r-bit binary hash value Hash for E using the hash function H3. XOR Hash with the r-bit binary number stored in the cell at index1 of L1 and index2 of L2 respectively. The result of the XOR replaces the original r-bit binary number in the corresponding cell.

[0018] Furthermore, the contents of the two strings L1 and L2 at the current time are copied and preserved as snapshots; in order to cover all the necessary historical information, the values ​​of time points T1 and T2 should be slightly earlier than the start time and slightly later than the end time of the required time period, respectively.

[0019] Furthermore, the inconsistencies caused by timing issues are removed through the following steps:

[0020] (1) Remove elements d from the inconsistent set Set i The key and value are concatenated to obtain a new element E. Hash values ​​are calculated on E using hash functions H1 and H2. The remainder is taken with respect to the number of cells m in the string to obtain the corresponding cell position indices Index1 and Index2.

[0021] (2) Calculate a binary hash value Hash of size r bits using the hash function H3. XOR Hash with the r-bit binary number stored in the cell at index1 of C1 and index2 of C2 respectively to obtain two binary numbers D1 and D2.

[0022] (3) If at least one of D1 and D2 is 0, then replace the r-bit binary number in the cell at index1 of C1 with D1, replace the r-bit binary number in the cell at index2 of C2 with D2, and change the value of element d at this time. i Remove from Set and add to the historical information set H;

[0023] (4) Repeatedly traverse all elements in the Set until the number of elements in the Set no longer decreases. At this time, H is the set of all historical information from time point T1 to time point T2 in the Set. The remaining elements in the Set are not historical information.

[0024] Furthermore, if C1 and C2 are two strings of all zeros after step (4), it proves that all historical information between time point T1 and time point T2 has been restored and is considered to be fully decoded.

[0025] Furthermore, when addressing timing issues in network configuration consistency verification, the elements in the historical information set H and the inconsistency set Set are further processed based on the configuration type:

[0026] Add / Delete: The corresponding element can be found in H;

[0027] Update: Information updated between time point T1 and time point T2 can be found, but the updated content will be treated as inconsistent and left in the Set. At this time, there exists one and only one element in H and Set respectively, with the same key but different value. By deleting the element in Set with the same key as the element in H, the earlier version before the update is excluded from the inconsistency caused by the configuration update.

[0028] Furthermore, if the information to be configured is a long string, the entire long string can be hashed in the hash calculation step; this method is not limited to the format of configuration or other historical information that needs to be saved, as long as the format of the information is consistent throughout the process.

[0029] Compared with the prior art, the present invention has the following advantages:

[0030] First, the method has high reliability. This method, with minimal resource overhead, can achieve a reliability of less than 10... -7 The false negative probability data can be restored with a high probability, even when historical information is missing.

[0031] Second, it has low resource consumption. This method requires no additional storage resources, only 2*m*r bits of memory and a very small amount of computing resources to restore historical information;

[0032] Third, it has high sensitivity. This method can reflect historical information in real time over any finite period of time. For example, in consistency verification, this method can reduce the dwell time of faults to within a verification cycle.

[0033] Fourth, the method is robust. This method only retains snapshots of two time points, and the process of recording configuration information does not involve the internal functional implementation of network elements. Therefore, it can be regarded as a stateless network function that requires no additional maintenance. Attached Figure Description

[0034] Figure 1 This is a schematic diagram of a network configuration history information storage method provided in an embodiment of the present invention;

[0035] Figure 2 This refers to parameter selection under the network configuration consistency verification scenario provided in the embodiments of the present invention;

[0036] Figure 3 This demonstrates the method performance of the embodiments of the present invention under suitable parameters;

[0037] Figure 4 The results are the stability test results of the method under appropriate parameters in the embodiments of the present invention. Detailed Implementation

[0038] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0039] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0040] This embodiment provides specific implementation details of the method in the scenario of consistency verification of cloud network configuration:

[0041] like Figure 1 As shown in the example, this application provides a method for summarizing configuration history information of cloud network element devices. Two long strings, L1 and L2, each consisting of m r-bit cells, are created in the memory of the network element device. First, the network configuration type is defined as three categories: update, delete, and add. The configuration information is distributed in key-value pair format. The specific content of the three types of configurations is as follows:

[0042] 1. Update: Given the contents of a key-value pair, keep the key the same and change the value.

[0043] 2. Delete: Given the contents of a key-value pair, delete the corresponding contents;

[0044] 3. Added: Given the content of a key-value pair, add the corresponding content.

[0045] Generally, to verify the configuration received from the controller on the network element device and the actual effective configuration, the network element device will update the status of the received configuration in its local database according to its type (i.e., deletion, update, and addition), thereby obtaining a verification reference for the effective configuration on the network element device. Based on this, this embodiment records configuration summary information through the following process:

[0046] 1. Concatenate the key and value of the configuration received from the controller to obtain a long string E. Calculate a hash value for E using hash functions H1 and H2 respectively. Take the remainder after dividing by the number of cells m in strings L1 and L2 to obtain the corresponding cell indexes Index1 and Index2. The concatenation method could be simple string addition. The reason for concatenating configurations is that different configurations may have the same key or value, leading to identical results when calculating using only the key and value, potentially causing false positives beyond hash collisions. While using the concatenated result for hash calculation increases computational overhead due to the longer string, it ensures completely random hash results for different configurations. Furthermore, since the number of configurations processed per second generally does not exceed 5000, the computational resources required for processing are limited, and any additional resource allocation is also quite limited.

[0047] 2. Calculate an r-bit binary hash value Hash using the hash function H3 on E. XOR Hash with the r-bit binary number stored in the cell at index1 of L1 and index2 of L2 respectively. Replace the original r-bit binary number in the corresponding cell with the result of the XOR.

[0048] The process of storing configuration history information in L1 and L2 on network element devices is synchronized with the process of the controller issuing configuration to network element devices. Configuration history information within a certain period is restored based on snapshot information between two defined time points. For inconsistencies found when performing network configuration consistency checks on the network element device between the historical configuration received from the controller and the actually effective configuration, these inconsistencies fall into two categories: one is inconsistency caused by time synchronization issues, and the other is inconsistency caused by actual configuration errors. In this case, it is necessary to use the configuration history information to delete the inconsistencies caused by time synchronization issues. The specific process is as follows:

[0049] 1. Before initiating the network configuration consistency verification process, this time is recorded as time point T1. First, save snapshots S1 of L1 and L2.

[0050] 2. Extract the actual applied configuration entries from the backend program daemon of the network element device, and extract the verification reference of the effective configuration from the local agent program.

[0051] 3. After obtaining the two configuration sets that need to be verified for network configuration consistency, this point is recorded as time point T2, and snapshots S2 of L1 and L2 are saved.

[0052] For the inconsistencies found in network configuration consistency verification, the portion belonging to the actual applied configuration entries is denoted as the inconsistency set DIFF1, and the portion belonging to the effective configuration verification reference is denoted as the inconsistency set DIFF2. The union of DIFF1 and DIFF2 is denoted as Set. This method is then used to remove inconsistencies caused by timing issues. The specific process is as follows:

[0053] 1. XOR S1 with L1 and L2 in S2 respectively to obtain two long strings C1 and C2.

[0054] 2. Add element d from the Set to be confirmed. i The key and value are concatenated to obtain a new element E. Hash values ​​are calculated on E using hash functions H1 and H2 respectively. The remainder is taken with respect to the number of cells m in the string to obtain the corresponding cell position indices Index1 and Index2.

[0055] 3. Calculate a binary hash value Hash of size r bits using the hash function H3. XOR Hash with the r-bit binary number stored in the cell at index1 of C1 and index2 of C2 respectively to obtain two binary numbers D1 and D2.

[0056] 4. If at least one of D1 and D2 is 0, it indicates that the inconsistency is due to a timing issue. Replace the r-bit binary number in cell C1 at index 1 with D1, and replace the r-bit binary number in cell C2 at index 2 with D2. In other words, delete the element corresponding to E from C1 and C2, and then change the current element d. i Remove it from the Set and add it to the historical information set H.

[0057] 5. Repeatedly traverse all elements in the Set until the number of elements in the Set no longer decreases. At this point, H is the set of all historical information from time point T1 to time point T2 in the Set. Furthermore, for inconsistencies caused by configuration updates, by deleting elements in the Set with the same key as elements in H, the earlier version before the update can be excluded from the inconsistencies caused by configuration updates. For configuration verification scenarios in cloud networks, this embodiment can efficiently restore historical configuration information based on the record summary of historical information, given the inconsistencies caused by actual configuration errors.

[0058] To further consider the accuracy of this process, we need to examine both false positives and false negatives:

[0059] 1. False Positive Rate (FPR): A false positive refers to a portion of non-historical information being treated as historical information. This corresponds to the situation where, after calculating D1 and D2 using the hash function H3, at least one value is 0 due to a hash collision. For a cell with position r, the probability can be expressed as:

[0060]

[0061] When r is 32, the corresponding false positive rate is 4.66 * 10^32. -10 It can be ignored.

[0062] 2. False Negative Rate (FNR): This refers to the recall aspect, corresponding to the method's ability to reconstruct historical information. In fact, decoding failure occurs only when the positions of historical information mapped to strings happen to overlap. For example, for E1 and E2 obtained by concatenating the key and value of elements d1 and d2, the calculated H1(E1) = H1(E2), H2(E1) = H2(E2). Furthermore, for E1, E2, E3, and E4, we have H1(E1) = H1(E2) = H1(E3) = H1(E4), H2(E3) = H2(E1) = H2(E2) = H2(E4), or H1(E1) = H1(E4), H2(E1) = H2(E2), H1(E3) = H1(E2), H2(E1) = H2(E2), and so on. Therefore, overlapping elements will interfere with each other during decoding. Any overlapping element A has a mapping that includes a mapping to another element B, which is itself interfered with by other elements. This creates a closed loop between the interfering elements, making decoding impossible. When any number of elements overlap, the decoding attempt results in a false negative. Using classical probability models and the partitioning principle, the upper bound of the probability of any number of overlapping elements after inserting n elements into a string consisting of m cells can be estimated as:

[0063]

[0064] Therefore, the false negative rate can be roughly estimated as follows:

[0065]

[0066] Take the length of m as 2 20 When inserting 200,000 elements, the probability of a false negative is less than 2 * 10. -7 .

[0067] It can be observed that the performance of this method is directly related to the number of cells *m* and the bit size *r* of the cells in strings *L1* and *L2*. Since the register size on network element devices is generally 64 bits, this embodiment chooses 32 bits as the cell bit size. At this point, false positives are essentially negligible, while simultaneously improving the efficiency of hash calculation. Regarding the number of cells *m*, as... Figure 2 As shown, based on the following characteristics:

[0068] 1. The false negative rate of the method should be as low as possible;

[0069] 2. The method should have the strongest possible fault tolerance capability;

[0070] 3. The parameter selection should be suitable for hash calculations, i.e., powers of 2;

[0071] In this embodiment, the size of m is chosen to be 2. 20 Based on this, the upper bound of the false negative rate calculated by this method was tested using the Monte Carlo experiment, such as... Figure 3 As shown, the experimental results are consistent with the expected calculations; when inserting 20,000 elements, the false negative rate does not exceed 2*10. -7 .

[0072] Finally, due to frequent changes in network configuration, some configurations, although not overlapping in C1 and C2, may not be fully decoded from both C1 and C2. For example, if the same key is updated more than twice within a short period, the inconsistency detected by consistency checks will only include the latest version. Versions updated before this time period are normally mapped to strings, but because they are no longer in the Set, they cannot be decoded. The positions mapped to these elements are equivalent to the overlapping positions of the elements; if an element is mapped to the same position twice, it also cannot be decoded. This method defines the duty cycle as the proportion of historical information in the Set to all historical information in T1 to T2. When the duty cycle is 1, the Set contains all historical information from T1 to T2. Based on different duty cycles and different numbers of inserted elements, this method further tests its performance using Monte Carlo experiments. Figure 4 As can be seen, the method's performance remains stable even with a large duty cycle, which aligns with the data characteristics and requirements of real-world scenarios. Furthermore, even if some information cannot be decoded from this method, treating a small number of timing inconsistencies as misconfigurations during configuration consistency verification will not significantly burden system performance.

[0073] This method was tested in an Alibaba Cloud test environment to examine the overhead of storing historical information. When storing 10,000 configuration records per second with an average length of 330 bytes, the CPU overhead required in a single-threaded environment was approximately 30ms, which is negligible. Furthermore, when m=2... 20When r=32, the required additional memory overhead is only 8MB. Using this method, the system can confirm whether a configuration belongs to historical information within a certain period without additional storage overhead. Under normal production conditions, the recall rate for restoring a limited historical information exceeds 99.99999%. This method effectively improves the sensitivity of configuration consistency verification, ensuring that faults are detected within a reconciliation cycle, thereby minimizing fault latency and improving network service quality.

[0074] This method is applicable to any scenario requiring the preservation of historical information, and is not limited to cloud network configurations. The above description is merely a preferred embodiment of the present invention. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the invention. Any person skilled in the art can make many possible variations and modifications to the technical solutions of the present invention using the disclosed methods and techniques, or modify them into equivalent embodiments with equivalent changes, without departing from the scope of the present invention. Therefore, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention, without departing from the content of the present invention, still fall within the protection scope of the present invention.

Claims

1. A highly efficient method for processing cloud network configuration information, characterized in that, Includes the following steps: S1, create two strings L1 and L2, each consisting of m r-bit cells, in the network element's memory; whenever the network element receives network configuration information from the controller, map the binary hash value calculated using a hash function to strings L1 and L2, XOR it with the original information at the corresponding position, and then store it in the corresponding position of the string, specifically: (1) Concatenate the key and value of the network configuration information received from the controller to obtain string E. Calculate a hash value for E using hash functions H1 and H2 respectively. Take the remainder of the number of cells m in strings L1 and L2 to obtain the corresponding cell position indices Index1 and Index2. (2) Calculate an r-bit binary hash value Hash for E using the hash function H3. XOR Hash with the r-bit binary number stored in the cell at index1 of L1 and index2 of L2 respectively. The result of the XOR replaces the original r-bit binary number in the corresponding cell. S2, obtain a summary of network configuration history information within a given time period, including: obtaining snapshots S1 and S2 of two strings L1 and L2 at two given time points T1 and T2, and XORing L1 and L2 in snapshots S1 and S2 at different time points respectively to obtain two strings C1 and C2; S3, obtain the set of inconsistencies between the network configuration information issued by the controller and the actual effective network configuration information of the network element device; S4, restore the network configuration history information within a given time period, including: for all elements in the inconsistent set Set, perform an XOR operation on the binary number recorded at the corresponding position in strings C1 and C2 and the binary hash value calculated by the hash function, and determine whether the corresponding element belongs to the network configuration history information within the given time period based on whether the result is 0; if the element belongs to the network configuration history information, remove the element from strings C1, C2 and set Set.

2. The cloud network configuration information processing method according to claim 1, characterized in that, The network configuration information includes three categories: update, deletion, and addition. The network configuration information is distributed in the form of key-value pairs. The specific content of the three types of network configuration information is as follows: Update: Given the content of a key-value pair, the key remains unchanged, but the value is changed; Delete: Given the contents of a key-value pair, delete the corresponding contents; Add: Given the content of a key-value pair, add the corresponding content; when the network configuration information is stored as a string, the configuration type is not recorded, only the key-value pair content is recorded.

3. The cloud network configuration information processing method according to claim 1, characterized in that, Copy the contents of the two strings L1 and L2 at the current time as a snapshot; in order to cover all the necessary historical information, the values ​​of time points T1 and T2 should be earlier than the start time and later than the end time of the required time period, respectively.

4. The cloud network configuration information processing method according to claim 1, characterized in that, Remove inconsistencies caused by timing issues using the following steps: (1) Remove elements d from the inconsistent set Set i The key and value are concatenated to obtain a new element E'. Hash values ​​are calculated on E' using hash functions H1 and H2 respectively. The remainder is taken with respect to the number of cells m in the string to obtain the corresponding cell position indices Index'1 and Index'2. (2) Calculate an r-bit binary hash value Hash' by hash function H3 on E'. XOR Hash' with the r-bit binary number stored in the cell at index'1 of C1 and index'2 of C2 respectively to obtain two binary numbers D1 and D2. (3) If at least one of D1 and D2 is 0, then replace the r-bit binary number in the cell at index'1 of C1 with D1, replace the r-bit binary number in the cell at index'2 of C2 with D2, and change the value of element d at this time. i Remove from Set and add to the historical information set H; (4) Repeatedly traverse all elements in the Set until the number of elements in the Set no longer decreases. At this point, H is the set of all historical information from time point T1 to time point T2 in the Set.

5. The cloud network configuration information processing method according to claim 4, characterized in that, If C1 and C2 are two strings of all zeros after step (4), it proves that all historical information from time point T1 to time point T2 has been restored and is considered to be fully decoded.

6. The cloud network configuration information processing method according to claim 4, characterized in that, When dealing with timing issues in network configuration information consistency verification, the elements in the historical information set H and the inconsistency set Set are further processed based on the configuration type: Add / Delete: The corresponding element can be found in H; Update: Information updated between time point T1 and time point T2 can be found, but the updated content will be treated as inconsistent and left in the Set. At this time, there exists one and only one element in H and Set respectively, with the same key but different value. By deleting the element in Set with the same key as the element in H, the earlier version before the update is excluded from the inconsistency caused by the configuration update.

7. The cloud network configuration information processing method according to claim 1, characterized in that, If the configuration information is a long string, hash the entire long string in the hash calculation step.

8. The cloud network configuration information processing method according to claim 1, characterized in that, For configuration totals in the tens of millions, set the number of cells m=2. 20 The number of digits in the cell is r=32.

Images