Server content updating and rollback method based on static technology

Abstract

The invention relates to the technical field of server content management, in particular to a server content updating and rollback method based on a static technology. According to the method, the change of the storage hot spots is detected through the entropy difference matrix, the characteristic moments are marked in combination with the timestamps, accurate division and unique identification of historical versions are achieved, redundant operation of full-amount scanning is avoided, versioning processing is only carried out on the change hot spot regions, the storage space utilization rate is remarkably optimized, and the version tracing efficiency and accuracy are improved; according to the method, the incremental patch is generated by adopting the binary differential algorithm accelerated by the suffix array, the optimal transmission path is constructed by combining network topology optimization, and the erasure code check information is added for the data block, so that the data transmission quantity is reduced, and meanwhile, the transmission fault tolerance is enhanced through the erasure code; and the data integrity and the transmission efficiency during network fluctuation or node failure are ensured.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of server content management, and in particular to a server content updating and rollback method based on static technology. BACKGROUND

[0002] Server content management technology covers multiple links such as content storage, distribution, version control, security guarantee and intelligent scheduling, aiming to improve content service efficiency, storage utilization and user experience. Server rollback is a core operation in server content management, which refers to the process of restoring the server state (including files, databases, configurations, etc.) to a historical version when system failure, data error or configuration exception occurs. The core goal is to quickly restore service availability and reduce business interruption time.

[0003] Firstly, the version control of the existing server rollback system relies on full file scanning, which cannot accurately identify the change area, resulting in waste of storage space.

[0004] Secondly, the existing server rollback system is based on timestamp or manual marking version management (such as Git), lacks dynamic hot area detection, and is difficult to accurately divide historical versions, with low traceability efficiency. Without combining network topology optimization, the transmission path is not optimal, with high delay.

[0005] In view of the above problems, it is necessary to propose a server content updating and rollback method based on static technology. SUMMARY

[0006] The purpose of the present application is to solve the problems in the background art and propose a server content updating and rollback method based on static technology.

[0007] The purpose of the present application can be achieved by the following technical solutions: A server content updating and rollback method based on static technology, comprising the following steps: Step 1, version metadata acquisition and entropy detection; Access the storage module of the server, perform file unit block division, and divide the static content into fixed-size data blocks. Calculate the byte-level information and entropy distribution data of each data block, and detect the change hotspot area based on the entropy value difference matrix. Perform data fusion on the byte-level information and entropy distribution data of all data blocks through cascading hash operation, to generate the version unique identifier and version fingerprint of the server static content.

[0008] Access the storage module of the server every preset time interval, scan the static file set and divide it by block to obtain a plurality of data blocks with a fixed size of 4MB.

[0009] The information entropy of all data blocks is calculated, the frequency Pk of each byte value k in each data block is counted, and the preset formula The information entropy E(Bi) of each data block Bi is calculated, wherein i is the data block number, i=1, 2,..., n; and n is the total number of data blocks.

[0010] As a preferred mode of the present application, the version fingerprint of the server is generated by a fingerprint generator, and the specific process is as follows: The information entropy of all data blocks is concatenated to obtain a combined entropy value {B1|B2|...|Bn}, and the combined entropy value is subjected to SHA3-512 operation to obtain the version fingerprint Vf(t) of the server at the current time, wherein t is a time stamp, i.e., the generation time of the version fingerprint Vf(t).

[0011] As a preferred mode of the present application, an entropy matrix is constructed based on the information entropy, the information entropy E(Bi) of each data block i is arranged on the diagonal line, and the difference between the information entropies E(Bi) is arranged in other positions to obtain an entropy matrix at the current time t: ,..., ,..., is an entropy difference, representing the difference between the information entropies.

[0012] represents the difference between the information entropy E(B1) of the data block 1 and the information entropy E(B2) of the data block 2.

[0013] The standard deviation σ(E, t) of all entropy differences is calculated, which is used to identify the change of the hot area of the storage space of the server.

[0014] When the absolute value of the difference between the standard deviation σ(E, t) of the entropy difference at the current time and the standard deviation σ(E, t-1) of the entropy difference at the previous time is greater than a preset threshold value, it is determined that the change of the hot area of the storage space of the server has occurred between the t-1 and t moments, and the t moment is marked as a feature moment.

[0015] The historical versions of the server storage module are divided according to the feature moments, and all data records stored by the server storage module between two adjacent feature moments are recorded as a historical version.

[0016] All historical versions are numbered, and the number symbol is j, j=1, 2,..., m; and m is the total number of historical versions.

[0017] Step two, incremental encoding and topology optimization; Constructing the incremental data package and distribution path, generating binary patch based on version fingerprint difference; calculating the optimal transmission path according to node topology, and adding error correction code check information to the data block. All incremental data packages are encapsulated as atomic transaction data units as the minimum operation unit for subsequent version rollback and version switching, and the network-aware minimum transmission coding is completed.

[0018] Obtaining the version fingerprint Vf(t) of each time t, the version fingerprint Vf(t-1) of the previous time t-1 and the entropy value matrix E(t).

[0019] Based on the entropy value matrix, the binary difference algorithm accelerated by suffix array is used to calculate the entropy value matrix encoding characteristic value The longest common subsequence of the new and old files is found by using suffix sorting, the data block that is not modified is marked, and the data block number series that is not modified compared with the previous time is obtained.

[0020] As a preferred mode of the application, the server node number e where each data block i is located is obtained, and the real-time network delay and available bandwidth are obtained. The optimal distribution tree decision model is constructed: Solving the node set T under the minimum condition , wherein e is the server node number, , and are the network delay, available bandwidth and maximum hop number of the distribution book of node e, respectively, wherein λ is a preset weight factor, and the default value is 0.7. When the value of λ is 1, it represents reducing intermediate nodes and reducing failure rate; when the value of λ is 0, it represents preferentially selecting short-hop high-speed link.

[0021] The version fingerprint Vf(t) of the server is assigned with public key and private key. The version fingerprint Vf(t), the public key and the private key of the version fingerprint, the entropy value matrix encoding characteristic value and the node set T are encapsulated as transaction messages, which are directly entered into the version arbitrator memory pool through RDMA, and are used as the basic data for hardware arbitration and version switching decision.

[0022] Step three, hardware arbitration and version switching decision; Based on the incremental data package, the abnormality is detected, the cryptographic signature of each data package is verified, the stability index of the historical version is calculated, the target version of the optimal rollback operation is determined based on the system state transition probability, and the rollback trigger signal is output.

[0023] The entropy value matrix encoding characteristic value and the node set T are analyzed by the version arbitrator, the version fingerprint is verified, and the historical version stability score is calculated, and the specific process is: The fingerprint generator is re-called to calculate the real-time version fingerprint Vf(t) of the server, the encapsulated transaction message is obtained from the version arbitrator memory pool, and the public key of the version fingerprint Vf(t) of the current time server is obtained. The real-time version fingerprint is verified by the public key of the version fingerprint of the current time server, and if the verification fails, a server information tampering signal is output.

[0024] The duration of each historical version Aj, the number of faults Fj and the number of data blocks contained in the data block number series of the historical version in which no data modification has occurred Nj are obtained, and the stability score Sj of each historical version j is obtained by weighted summation of Aj, Fj and Nj.

[0025] As a preferred mode of the present application, dynamic switching decision analysis is performed by Markov decision process, and when the server information tampering signal is identified, the following three preset server maintenance operations are obtained: Operation 1: Keep the current version; Operation 2: Switch to the version with the highest stability score of the historical version; Operation 3: Switch to the last version; The expected value of actions 1, 2 and 3 is calculated by the following formula: , and where a=1, 2, 3; corresponds to the three server maintenance operations, i.e. keeping the current version, switching to the version with the highest stability score of the historical version and switching to the last version.

[0026] where a is the encapsulated transaction message, i.e. the version fingerprint Vf(t), the public and private keys of the version fingerprint, and the entropy matrix encoding feature value and the node set T represents the current state of the server.

[0027] where is the immediate reward, i.e. when the last historical encapsulated transaction message is , the stability score rises by a after taking action a, where γ is a preset weight factor. Where is a historical transaction message similar to the current transaction message s, i.e. and the information entropy difference is less than a preset threshold.

[0028] where is the state transition probability, representing the probability of taking action a=1, 2, 3 in the historical transaction message .

[0029] where​ the reference value of the historical transaction message the similarity between the current transaction message s and the historical transaction message , and the difference between the information entropy of s and s.

[0030] the maximum value among the expected values of actions 1, 2 and 3.

[0031] if the maximum value is 1, output a maintenance signal; if the maximum value is 2, output a rollback trigger signal, and locate the target version of the rollback operation as the historical version with the highest stability score; if the maximum value is 3, output a rollback trigger signal, and locate the target version of the rollback operation as the previous historical version.

[0032] Step four, atomic version switching and rollback execution; After receiving the rollback trigger, locate the target version of the optimal rollback operation, execute the rollback operation, and complete the final execution of the version switching operation, including updating the storage mapping through the compare-and-swap instruction, redirecting the service access logical address, and log rollback. After receiving the atomic switching instruction, execute data restoration in units of data blocks.

[0033] After receiving the maintenance signal, no additional operation is taken; After receiving the rollback trigger signal, locate the target version of the rollback operation, obtain the information contained in all data blocks in the target version and the server node number e where the information is stored. Establish a storage mapping {i→e} of the data blocks to the server node number e, and perform location migration and storage content restoration on the existing data blocks. Record the operation type and starting time of the rollback operation, and record it in the rollback log.

[0034] As a preferred mode of the present application, if the target version of the rollback operation is the previous version, the calculation result of the binary difference algorithm accelerated by the suffix array is called to obtain the data restoration patch from the current version to the previous version.

[0035] Compared with the prior art, the present application has the following beneficial effects: 1. The present application realizes accurate division and unique identification of historical versions by detecting storage hot area changes through entropy difference matrix and combining time stamp marking feature moments. This technology avoids the redundant operation of full-scan, only processes versioning for the changed hot area, significantly optimizes storage space utilization, and improves the efficiency and accuracy of version tracing; ​​​​​​​2、The application adopts the binary differential algorithm accelerated by the suffix array to generate the incremental patch, combines the network topology optimization to build the optimal transmission path, and adds the erasure code check information to the data block. This design greatly reduces the data transmission amount, simultaneously enhances the transmission fault tolerance through the erasure code, and ensures the data integrity and transmission efficiency when the network fluctuates or the node fails; 3、The application dynamically evaluates the stability of the historical version, combines the cryptography signature verification to ensure the version authenticity. The mechanism can automatically select the optimal rollback target, and guarantees the atomicity and data consistency of the rollback process through the atomization operation, significantly improves the tamper resistance and fault recovery efficiency of the system. BRIEF DESCRIPTION OF DRAWINGS

[0036] In order to facilitate the understanding of those skilled in the art, the technical solutions of the application will be further described below with reference to the drawings: Figure 1 The method flowchart of the application. DETAILED DESCRIPTION

[0037] The technical solutions of the application will be described below in detail with reference to the embodiments. Obviously, the described embodiments are only a part of the embodiments of the application, but not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative work belong to the protection scope of the application.

[0038] Please refer to Figure 1 The server content update and rollback method based on static technology shown in the figure includes the following steps: Step one, versioned metadata acquisition and entropy detection; Access the storage module of the server, perform file unit block, divide the static content into data blocks with fixed size. Calculate the byte level information and entropy distribution data of each data block, and detect the change hot area based on the entropy value difference matrix. Through the cascade hash operation, the byte level information and entropy distribution data of all data blocks are fused to generate the version unique identifier and version fingerprint of the server static content.

[0039] Every preset time interval, access the storage module of the server, scan the static file set and divide by block to obtain several data blocks with fixed size of 4MB.

[0040] Calculate the information entropy of all data blocks, count the frequency Pk of each byte value k in each data block, and calculate the information entropy E(Bi) of each data block Bi through the preset formula The information entropy E(Bi) of each data block Bi is calculated, wherein i is the data block number, i=1,2,...,n; n is the total number of data blocks.

[0041] Further, the version fingerprint of the server is generated by the fingerprint generator, and the specific process is as follows: The information entropy of all data blocks is concatenated to obtain a combined entropy value {B1|B2|...|Bn}; the combined entropy value is subjected to SHA3-512 operation to obtain the version fingerprint Vf(t) of the server at the current time, wherein t is a time stamp, i.e., the generation time of the version fingerprint Vf(t).

[0042] Further, the entropy value matrix is constructed based on the information entropy, the information entropy E(Bi) of each data block i is arranged on the diagonal line, and the difference between the information entropies E(Bi) is arranged in other positions to obtain the entropy value matrix at the current time t: ,..., ,..., is an entropy value difference, representing the difference between the information entropies.

[0043] represents the difference between the information entropy E(B1) of the data block 1 and the information entropy E(B2) of the data block 2.

[0044] The standard deviation σ(E,t) of all entropy value differences is calculated, which is used to identify the change of the hot area of the storage space in the server.

[0045] When the absolute value of the difference between the standard deviation σ(E,t) of the entropy value difference at the current time and the standard deviation σ(E,t-1) of the entropy value difference at the previous time is greater than a preset threshold value, it is determined that the change of the hot area of the storage space in the server has occurred between t-1 and t, and the time t is marked as a feature time.

[0046] The historical versions of the server storage module are divided according to the feature time, and all data records stored in the server storage module between two adjacent feature times are recorded as a historical version.

[0047] All historical versions are numbered, and the number is j, j=1, 2,..., m; m is the total number of historical versions.

[0048] Step two, incremental encoding and topology optimization; The incremental data packet and the distribution path are constructed, the binary patch is generated based on the version fingerprint difference; the optimal transmission path is calculated according to the node topology, and the error correction code check information is added to the data block. All incremental data packets are encapsulated as atomic transaction data units as the minimum operation unit for subsequent version rollback and version switching, and the network-aware minimized transmission encoding is completed.

[0049] The version fingerprint Vf(t) at each time t, the version fingerprint Vf(t-1) at the previous time t-1, and the entropy value matrix E(t) are obtained.

[0050] Difference encoding based on entropy matrix, using suffix array accelerated binary difference algorithm to calculate the entropy matrix encoding characteristic value The longest common subsequence of the new and old files is found using suffix sorting, and the data blocks that are not modified are marked to obtain the data block sequence that has not been modified compared with the previous time.

[0051] It should be noted that bsdiff is an incremental update algorithm for binary files, specifically designed for efficient generation of file difference patches.

[0052] Further, the server node number e where each data block i is located is obtained, and its real-time network delay and available bandwidth are obtained. An optimal distribution tree decision model is constructed: Solve the node set T that satisfies The minimum condition, where e is the server node number, 、 and are the network delay, available bandwidth and maximum hop count of node e, respectively, where λ is a preset weight factor, and the default value is 0.7. When adjusting the value of λ to 1, it represents reducing intermediate nodes and reducing failure rate; when adjusting the value of λ to 0, it represents preferentially selecting short-hop high-speed links.

[0053] Assign public and private keys to the server's version fingerprint Vf(t). Encapsulate the version fingerprint Vf(t), the public and private keys of the version fingerprint, and the entropy matrix encoding characteristic value With the node set T into a transaction message, and through RDMA directly into the version arbitrator memory pool as the basic data for hardware arbitration and version switching decision.

[0054] Step three, hardware arbitration and version switching decision; Based on the incremental data packet, the abnormality is detected, the cryptographic signature of each data packet is verified, the stability index of the historical version is calculated, the optimal rollback operation target version is determined based on the system state transition probability, and the rollback trigger signal is output.

[0055] Through the version arbitrator, the entropy matrix encoding characteristic value And the node set T are analyzed, the version fingerprint is verified, and the historical version stability score is calculated, the specific process is: Recall the fingerprint generator to calculate the real-time version fingerprint Vf(t) of the server, obtain the encapsulated transaction message from the version arbitrator memory pool, and obtain the public key of the version fingerprint Vf(t) of the server at the current time. The real-time version fingerprint is verified through the public key of the version fingerprint of the server at the current time, and if the verification fails, the server information tampering signal is output.

[0056] The running time Aj, the number of failures Fj and the number of data blocks Nj contained in the data blocks in which no data modification has occurred in the historical version j are obtained, and the stability score Sj of each historical version j is obtained by weighted summation of Aj, Fj and Nj.

[0057] Further, dynamic switching decision analysis is performed by Markov decision process, and when the server information tampering signal is identified, the following three preset server maintenance operations are obtained: Operation 1: Keep the current version; Operation 2: Switch to the version with the highest stability score of the historical versions; Operation 3: Switch to the previous version; The expected values of actions 1, 2 and 3 are calculated by the following formula: The expected values of actions 1, 2 and 3 are calculated by the following formula: wherein a = 1, 2, 3; corresponding to the three server maintenance operations, i.e. keeping the current version, switching to the version with the highest stability score of the historical versions and switching to the previous version.

[0058] wherein a is the encapsulated transaction message, i.e. version fingerprint Vf(t), public and private keys of the version fingerprint, and entropy matrix encoding feature value and node set T, representing the current state of the server.

[0059] wherein is the immediate reward, i.e. when the last historical encapsulated transaction message is , the increase in stability score after taking action a, wherein γ is a preset weight factor. Wherein is the historical transaction message similar to the current transaction message s, i.e. and The difference in information entropy is less than a preset threshold.

[0060] wherein is the state transition probability, representing the probability of taking action a = 1, 2, 3 in the historical transaction message .

[0061] wherein is the state value, representing the reference value of the historical transaction message , i.e. the similarity between the historical transaction message and the current transaction message s, and the specific value is The difference in information entropy between and s.

[0062] The expected values of actions 1, 2 and 3 are calculated by the following formula: ​​, and the maximum value in.

[0063] if the maximum value is , output a maintenance signal; if the maximum value is , output a rollback trigger signal and locate the target version of the rollback operation as the historical version with the highest stability score; if the maximum value is , output a rollback trigger signal and locate the target version of the rollback operation as the last historical version.

[0064] Step four, atomic version switching and rollback execution; After receiving the rollback trigger, locate the target version of the optimal rollback operation, execute the rollback operation, and complete the final execution of the version switching operation, including updating the storage mapping through the compare-and-swap instruction, redirecting service access to the logical address, and log rollback. After receiving the atomic switching instruction, execute data restoration in units of data blocks.

[0065] After receiving the maintenance signal, no redundant operation is taken; After receiving the rollback trigger signal, locate the target version of the rollback operation, obtain the information contained in all data blocks in the target version and the server node number e where it is stored. Establish a storage mapping {i→e} of data blocks to server node numbers e, and perform location migration and storage content restoration on the existing data blocks. Record the operation type and departure time of the rollback operation that occurs, and record it in the rollback log.

[0066] If the target version of the rollback operation is the last version, retrieve the calculation result of the binary difference algorithm accelerated by the suffix array to obtain the data restoration patch from the current version to the last version.

[0067] It should be understood that the terms "comprise" and "comprising" used in the specification and claims of the present disclosure indicate the presence of the described features, integers, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0068] It should also be understood that the terms used in the present disclosure specification are only for the purpose of describing specific embodiments and are not intended to limit the present disclosure. As used in the present disclosure specification and claims, the singular forms "a", "an" and "the" are intended to include the plural forms unless the context clearly indicates otherwise. It should be further understood that the term "and / or" used in the present disclosure specification and claims means any combination of one or more of the associated listed items and all possible combinations thereof; The preferred embodiments of the application disclosed above are only to facilitate the elucidation of the application. The preferred embodiments do not describe all the details of the application and limit the application to the specific embodiments. Obviously, many modifications and variations can be made in light of the teachings above. The description is chosen and described in order to provide the best illustration of the application and its practical application to those skilled in the art and to enable those skilled in the art to best utilize the application. The application is limited only by the claims and their full scope and equivalents.

Claims

1. A method for updating and rolling back server content based on static technology, characterized in that, Includes the following steps: Step 1: Versioned metadata collection and entropy detection; Access the server's storage module, perform file unit chunking, and divide static content into fixed-size data blocks; Calculate the byte-level information and entropy distribution data of each data block, and detect change hotspots based on the entropy difference matrix; perform data fusion on the byte-level information and entropy distribution data of all data blocks through cascaded hash operation to generate a unique version identifier and version fingerprint of the server's static content; Step 2: Incremental coding and topology optimization; Construct incremental data packets and distribution paths, generate binary patches based on version fingerprint differences; calculate the optimal transmission path according to node topology, and add erasure coding verification information to data blocks; encapsulate all incremental data packets into atomic transaction data units, which serve as the smallest operation unit for subsequent version rollback and version switching, and complete network-aware minimal transmission coding. Step 3: Hardware arbitration and version switching decisions; Anomaly detection is performed based on incremental data packets, the cryptographic signature of each data packet is verified, the stability index of historical versions is calculated, the version fingerprint is verified, and the stability score of historical versions is calculated. The optimal target version for the rollback operation is determined based on the system state transition probability, and a rollback trigger signal is output. Step 4: Atomic version switching and rollback execution; Upon receiving a rollback trigger, locate the optimal target version for the rollback operation, execute the rollback operation, and complete the final execution of the version switching operation, including updating the storage mapping through comparison exchange instructions, redirecting the service access logical address, and rolling back the logs; upon receiving an atomic switch instruction, perform data restoration on a block-by-block basis.

2. The server content update and rollback method based on static technology according to claim 1, characterized in that, The specific process of dividing static content into fixed-size data blocks and calculating entropy distribution data is as follows: At preset time intervals, the server's storage module is accessed to scan the static file collection and divide it into blocks to obtain several data blocks with a fixed size of 4MB. Calculate the information entropy for all data blocks, and statistically analyze the frequency Pk of each byte value k within each data block using a preset formula. Calculate the information entropy E(Bi) of each data block Bi, where i is the data block number, i=1,2,...,n; and n is the total number of data blocks. Generate the server's version fingerprint using a fingerprint generator.

3. The server content update and rollback method based on static technology according to claim 2, characterized in that, The specific process of generating a server version fingerprint using a fingerprint generator is as follows: The information entropy of all data blocks is concatenated to obtain the combined entropy value {B1|B2|...|Bn}; the combined entropy value is subjected to SHA3-512 operation to obtain the server version fingerprint Vf(t) at the current time, where t is the timestamp, that is, the generation time of the version fingerprint Vf(t); Based on the information entropy, an entropy matrix is ​​constructed. The information entropy E(Bi) of each data block i is arranged on the diagonal, and the differences between the information entropy E(Bi) are arranged in other positions to obtain the entropy matrix at the current time t: in, , ..., The entropy difference represents the difference in information entropy. The difference between the information entropy E(B1) of data block 1 and the information entropy E(B2) of data block 2; Calculate the standard deviation σ(E,t) of all entropy differences to identify changes in hot zones of storage space in the server and to classify historical versions.

4. The server content update and rollback method based on static technology according to claim 3, characterized in that, The specific process for dividing historical versions is as follows: When the absolute value of the difference between the standard deviation σ(E,t) of the entropy difference at the current moment and the standard deviation σ(E,t-1) of the entropy difference at the previous moment is identified... If the value exceeds the preset threshold, it is determined that a change in the server storage space hot zone has occurred between time t-1 and time t, and time t is marked as the characteristic time. Based on the characteristic time, the historical versions of the server storage module are divided, and all the data stored in the server storage module between two adjacent characteristic times are recorded as one historical version. All historical versions are numbered with the identifier j, where j = 1, 2, ..., m; m is the total number of historical versions.

5. A method for updating and rolling back server content based on static technology according to claim 1, characterized in that, The specific process for generating binary patches based on version fingerprint differences is as follows: Obtain the version fingerprint Vf(t) at each time t, the version fingerprint Vf(t-1) at the previous time t-1, and the entropy matrix E(t). Difference encoding is performed based on the entropy matrix, and a binary difference algorithm accelerated by suffix arrays is used to calculate the entropy matrix encoded eigenvalues. Use suffix sorting to find the longest common subsequence of the old and new files, mark the unmodified data blocks, and obtain the sequence of data blocks that have not been modified compared with the previous time step.

6. The method for updating and rolling back server content based on static technology according to claim 1, characterized in that, The specific process for calculating the optimal transmission path is as follows: Obtain the server node number e where each data block i is located, and obtain its real-time network latency and available bandwidth; construct the optimal distribution tree decision model: Solve for the satisfaction Given the minimum set of nodes T, where e is the server node number. , and These represent the network latency, available bandwidth, and maximum hop count for distribution of node e, respectively. λ is a preset weighting factor with a default value of 0.

7. Adjusting λ to 1 reduces intermediate nodes and lowers the failure rate; adjusting λ to 0 prioritizes short-hop, high-speed links. Assign a public and private key to the server's version fingerprint Vf(t); encode the version fingerprint Vf(t), the public and private keys of the version fingerprint, and the entropy matrix into feature values. The node set T is encapsulated as a transaction message and directly entered into the version arbitrator memory pool via RDMA, serving as the basis for hardware arbitration and version switching decisions.

7. A method for updating and rolling back server content based on static technology according to claim 1, characterized in that, The specific process for verifying version fingerprints and calculating historical version stability scores is as follows: Recall the fingerprint generator, calculate the server's real-time version fingerprint Vf(t), retrieve the encapsulated transaction message from the version arbitrator memory pool, and obtain the public key of the server's current version fingerprint Vf(t). Verify the real-time version fingerprint using the public key of the server's current version fingerprint. If the verification fails, output a server information tampering signal. Obtain the continuous runtime Aj, the number of failures Fj, and the number of data blocks Nj contained in the data block column that has not undergone data modification in each historical version. The stability score Sj of each historical version j is obtained by weighted summing of Aj, Fj, and Nj. Dynamic switching decision analysis is performed using Markov decision processes.

8. A method for updating and rolling back server content based on static technology according to claim 7, characterized in that, The specific process of dynamic switching decision analysis using Markov decision processes is as follows: Get three preset server maintenance operations: Operation 1: Keep the current version; Operation 2: Switch to the version with the highest historical stability score; Operation 3: Switch to the previous version; By using a preset formula: Calculate the expected values ​​of actions 1, 2, and 3. , and , where a=1,2,3; correspond to three server maintenance operations, namely, maintaining the current version, switching to the version with the highest historical stability score, and switching to the previous version; Where 'a' represents the encapsulated transaction message, namely the version fingerprint Vf(t), the public and private keys of the version fingerprint, and the entropy matrix encoded feature values. The set of nodes T represents the current state of the server; Among them For immediate rewards, i.e., the previously encapsulated transaction message was... When, the increase in stability score after taking operation a, where γ is a preset weighting factor; where For historical transaction messages similar to the current transaction message s, i.e. and The difference in information entropy is less than a preset threshold; Among them The state transition probability represents historical transaction messages. In the context, the probability of taking action a = 1, 2, 3; Among them State value represents historical transaction messages. The reference value of historical information The similarity to the current transaction message s, specifically the numerical value is: The difference in information entropy between and s; Obtain the expected values ​​of actions 1, 2, and 3. , and The maximum value in.

9. A method for updating and rolling back server content based on static technology according to claim 1, characterized in that, The final execution process for completing the version switch operation is as follows: No further action is taken after receiving the sustain signal; Upon receiving a rollback trigger signal, locate the target version of the rollback operation, obtain the information contained in all data blocks in the target version and their stored server node number e; establish a storage mapping {i→e} from data blocks to server node number e, and perform location migration and storage content restoration on the existing data blocks; Record the operation type and start time of any rollback operation, and record them in the rollback log; If the target version of the rollback operation is the previous version, retrieve the calculation result of the binary difference algorithm accelerated by the suffix array to obtain the data restoration patch from the current version to the previous version.

Images