A cloud computing service inter-communication protocol optimization method and system

Abstract

This application provides a method and system for optimizing secure communication protocols between cloud computing services, relating to the field of secure communication technology in cloud computing. By establishing a trusted channel between a microservice and a local security proxy service, it achieves local authentication of the microservice identity and data transmission. The local security proxy service establishes a long-term secure connection with the external service and manages its session keys. Short-lived sub-session keys are distributed for data encryption between the local security proxy service and the microservice. When the microservice encrypts business data using the sub-session key and sends it to the local security proxy service, the proxy service decrypts it and re-encrypts it using the session key of the long-term secure connection before forwarding it. This effectively resists side-channel attacks and overcomes the dilemma of balancing performance and security in existing technologies, providing an efficient and secure communication solution for multi-tenant, high-concurrency cloud computing environments.

Description

Technical Field

[0001] This application relates to the field of secure communication technology in cloud computing, and more specifically, to a method and system for optimizing secure communication protocols between cloud computing services. Background Technology

[0002] In modern cloud computing multi-tenant hybrid deployment environments, services need to establish encrypted channels through key exchange to ensure security. However, in high-concurrency scenarios, a large number of microservices simultaneously initiating key exchanges can drastically consume encryption hardware (such as CPU integrated encryption instruction units) and high-speed memory resources—operations such as digital certificate loading and random number generation compete for shared computing power and cache / memory bandwidth. Resource contention leads to longer secure connection establishment times, slower service response, and reduced data transmission efficiency, especially affecting services with high-frequency connections or periodically updated credentials.

[0003] To alleviate performance pressure, operations teams often extend the security credential update cycle to reduce key exchange frequency and temporarily lower resource consumption. However, this introduces a more insidious security risk: shared hardware performing encryption can leak side-channel information (such as computation time and cache access patterns). Under the original short-cycle strategy, attackers find it difficult to accumulate enough data to deduce the key; with the extended cycle, malicious tenants on the same physical server can monitor side-channel signals for an extended period and crack the key through statistical analysis or machine learning. Once the key is leaked, attackers can decrypt communications or forge identities.

[0004] Ultimately, cloud platforms face a dilemma: maintaining short lifecycles results in insufficient performance, while extending lifecycles increases security risks. Existing security protocols and optimization methods cannot resolve the unique challenges posed by multi-tenant resource contention, operational strategy adjustments, and the intertwining of side-channel attacks. Summary of the Invention

[0005] This application provides a method and system for optimizing secure communication protocols between cloud computing services, aiming to address the challenges faced by traditional secure communication protocols in terms of performance and security under high-density, multi-tenant microservice hybrid deployment modes in cloud computing environments, particularly performance bottlenecks caused by resource contention and the risk of side-channel attacks introduced by extending credential update cycles.

[0006] On one hand, this application provides a method for optimizing secure communication protocols between cloud computing services. The method is executed by a local security proxy service deployed on a physical server hosting microservices, and includes: Establish a trust channel with the microservice, the trust channel being used for the microservice's authentication and data transmission; The local security proxy service performs a secure communication protocol handshake with external services to establish a long-term secure connection, and manages the session key of the long-term secure connection. Distribute short-lived sub-session keys, which are used for data encryption between the local security proxy service and the microservice; The microservice receives business data encrypted with a sub-session key through the trusted channel. The local security proxy service decrypts the business data, re-encrypts it using the session key of the long-term secure connection, and forwards it to the external service.

[0007] On the other hand, this application provides a cloud computing service inter-service secure communication protocol optimization system, the system comprising: The deployment module is used to deploy a local security proxy service on the physical server hosting the microservices and configure resource isolation for the local security proxy service to ensure that its operational stability and performance are not affected by resource contention of the microservices on the physical server; The trust channel establishment module establishes a trust channel between the local security proxy service and the microservice, and the trust channel is used for the authentication and data transmission of the microservice. The handshake and key management module is used for the local security proxy service to perform a secure communication protocol handshake with the external service to establish a long-term secure connection, and the local security proxy service manages the session key of the long-term secure connection. The sub-session key distribution module is used by the local security proxy service to distribute short-lifetime sub-session keys, which are used for data encryption between the local security proxy service and the microservice. The data processing module is used for the local security proxy service to receive business data encrypted by the microservice using a sub-session key through the trusted channel, the local security proxy service to decrypt the business data, re-encrypt the business data using the session key of a long-term secure connection, and forward it to the external service.

[0008] The cloud computing service inter-service secure communication protocol optimization method and system involved in this application effectively solves the performance and security challenges faced by traditional security mechanisms in the high-density, multi-tenant microservice hybrid deployment mode in the existing cloud computing environment by deploying a local security proxy service on the physical server carrying the microservice and having the service perform a series of optimization steps.

[0009] In summary, this application significantly improves the performance of communication between cloud computing services without sacrificing security by introducing local security proxy services, hierarchical key management (combining long-term session keys with short-lifetime sub-session keys), and local trust channel mechanisms. It also effectively resists side-channel attacks, overcomes the dilemma of balancing performance and security in existing technologies, and provides an efficient and secure communication solution for multi-tenant, high-concurrency cloud computing environments. Attached Figure Description

[0010] To illustrate this application more clearly, the accompanying drawings used in the embodiments will be briefly described below. Obviously, those skilled in the art can obtain other drawings based on these drawings without any creative effort.

[0011] Figure 1 The diagram above illustrates a flowchart of a method for optimizing a secure communication protocol between cloud computing services in an embodiment. Figure 2 The diagram above illustrates a schematic of the structure of a cloud computing service inter-service secure communication protocol optimization system in an embodiment.

[0012] Figure reference numerals: 100, Cloud computing service inter-service secure communication protocol optimization system; 10, Deployment module; 20, Trust channel establishment module; 30, Handshake and key management module; 40, Sub-session key distribution module; 50, Data processing module. Detailed Implementation

[0013] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0014] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0015] In modern cloud computing environments, traditional inter-service secure communication protocols face a performance-security dilemma when dealing with high-density, multi-tenant microservice hybrid deployments. The simultaneous key exchange by numerous microservices leads to severe contention for encryption hardware and high-speed memory resources, increasing the latency of secure connection establishment and reducing overall data transmission efficiency. To alleviate performance pressure, operations teams sometimes extend the update cycle of security credentials, but this inadvertently provides opportunities for side-channel attacks, transforming theoretical security risks into practical security vulnerabilities.

[0016] like Figure 1 The illustration shows a flowchart of a method for optimizing a secure communication protocol between cloud computing services, as exemplarily shown in an embodiment. This application proposes a method for optimizing a secure communication protocol between cloud computing services, which is executed by a local security proxy service deployed on a physical server hosting microservices, and includes: S10, Establish a trust channel with the microservice, the trust channel being used for the microservice's authentication and data transmission; Microservices refer to a software architecture style that breaks down a single application into a set of small, loosely coupled services, each running in its own process and communicating through lightweight mechanisms.

[0017] A local security proxy service is a dedicated service deployed on the same physical server as the microservice. Its main responsibility is to handle security matters related to the microservice's external communications, including authentication, key management, and data encryption / decryption.

[0018] A trusted channel is a secure communication link established between a local security agent service and a microservice to ensure the confidentiality and integrity of data transmission between the two and to perform authentication.

[0019] S20, perform a secure communication protocol handshake with the external service to establish a long-term secure connection, and the local security proxy service manages the session key of the long-term secure connection; External services refer to services that communicate with microservices and are deployed on other physical servers or networks.

[0020] Long-term secure connections refer to encrypted communication connections with a relatively long lifespan established between local security agent services and external services, with the session keys managed by the local security agent service.

[0021] S30, Distribute a short-lifetime sub-session key, the sub-session key being used for data encryption between the local security proxy service and the microservice; A subsession key is a short-lived key that is issued by the local security proxy service to the microservice and used for data encryption between the local security proxy service and the microservice.

[0022] S40, the microservice receives business data encrypted with a sub-session key through the trusted channel, the local security proxy service decrypts the business data, re-encrypts the business data using the session key of the long-term secure connection, and forwards it to the external service.

[0023] The proposed method for optimizing secure communication protocols between cloud computing services first establishes a trusted channel between the local security proxy service and the microservice. This trusted channel forms the basis for secure communication between the local security proxy service and the microservice, and is used for microservice authentication and data transmission. For example, the local security proxy service can mutually authenticate with the microservice using pre-configured shared keys or certificates, thereby establishing a secure local communication link. Alternatively, the local security proxy service can utilize secure inter-process communication (IPC) mechanisms provided by the operating system, such as named pipes or shared memory, and add additional encryption and authentication layers to ensure the security of the trusted channel.

[0024] Secondly, the local security proxy service performs a secure communication protocol handshake with the external service to establish a long-term secure connection and manage the session keys for that connection. This step aims to decouple the complex external secure communication handshake process from the microservices and handle it uniformly by the local security proxy service. For example, the local security proxy service can act as a TLS / SSL client or server, performing a standard TLS handshake with the external service to negotiate and establish a long-term, valid secure connection. During this process, the local security proxy service is responsible for generating, storing, and updating the session keys for the long-term secure connection. These session keys can be stored in the local security proxy service's secure memory area and protected by access control mechanisms.

[0025] Next, short-lived sub-session keys are distributed, which are used for data encryption between the local security proxy service and the microservices. To avoid the security risks and performance overhead caused by microservices directly handling long-lived session keys, the local security proxy service distributes a short-lived sub-session key to each microservice or each communication session. For example, the local security proxy service can periodically generate a batch of new sub-session keys and distribute them to microservices through previously established trusted channels. Microservices use these sub-session keys for encryption when transmitting data with the local security proxy service. Alternatively, when a microservice needs to send data, it can request a sub-session key from the local security proxy service, which then dynamically generates and distributes it based on the request.

[0026] Finally, the local security proxy service receives the business data encrypted by the microservice using the sub-session key through the trusted channel. The local security proxy service decrypts the business data, re-encrypts it using the session key of the long-term secure connection, and forwards it to the external service. When a microservice needs to send business data to an external service, it first encrypts the business data using the sub-session key obtained from the local security proxy service and sends it to the local security proxy service through the trusted channel. Upon receiving the encrypted data, the local security proxy service decrypts it using the corresponding sub-session key to obtain the original business data. Subsequently, the local security proxy service re-encrypts the business data using the session key of the long-term secure connection established with the external service and forwards it to the external service through the long-term secure connection.

[0027] Compared to existing technologies, the core innovation of this application lies in its hierarchical key management and proxy communication mechanism. Traditional solutions typically require each microservice to independently establish and maintain secure connections with external services. This can lead to severe resource contention and performance degradation in high-concurrency, multi-tenant environments, and extending the key lifecycle increases the risk of side-channel attacks. This application introduces a local secure proxy service to centrally handle the complexity and resource consumption of external secure communication, and uses short-lifecycle sub-session keys to protect local communication, thus achieving a better balance between performance and security. This design not only addresses the pain points of traditional solutions but also provides a more efficient and secure solution for inter-service communication in cloud computing, significantly improving the overall security protection capabilities and operational efficiency of the cloud platform.

[0028] In some embodiments, the step of establishing a trust channel with the microservice specifically includes: Inject credentials for local authentication into the microservice; The microservice uses the credentials to initiate authentication with the local security proxy service; Once authentication is successful, a secure connection based on a local inter-process communication mechanism is established as a trusted channel between the local security proxy service and the microservice. The local inter-process communication mechanism is a Unix domain socket.

[0029] The locally authenticated credentials can include, but are not limited to, digital certificates, API keys, shared keys, or tokens. Their purpose is to provide a trusted identity for microservices, enabling verification at the local security proxy service. The injection of these credentials can be performed during microservice startup via environment variables, configuration files, or a security configuration service, ensuring the secure distribution and storage of credentials.

[0030] When a microservice initiates authentication with the local security proxy service, it may employ a challenge-response mechanism, credential-based signature verification, or mutual TLS (Transport Layer Security) protocols. This authentication process aims to verify the legitimacy of the microservice and prevent unauthorized microservices from establishing connections with the local security proxy service.

[0031] Once authentication is successful, a secure connection based on a local inter-process communication (IPC) mechanism will be established between the local security proxy service and the microservice. This connection serves as a trusted channel dedicated to microservice authentication and data transmission. Local IPC mechanisms, such as Unix domain sockets, are chosen for this purpose because they provide an efficient, low-latency, and highly secure local communication method. Unix domain sockets allow processes on the same physical server to communicate via file system paths, avoiding the overhead of network protocol stacks and leveraging file system permissions for access control, thereby enhancing the security of local communication.

[0032] The technical solution of this application injects pre-configured local authentication credentials into microservices during startup or runtime, enabling microservices to initiate trusted identity authentication with a local security proxy service. Once the microservice's identity is verified by the local security proxy service, a secure connection is established based on a local inter-process communication mechanism (such as Unix domain sockets). This connection mechanism leverages the local communication optimizations and security features provided by the operating system, ensuring that data transmission between the microservice and the local security proxy service is not only efficient but also effectively resists attacks from external networks, as communication traffic does not need to leave the local host. In this way, the establishment of a trust channel is strictly controlled within the local environment, significantly reducing the attack surface during trust establishment.

[0033] Through the above technical solutions, this application provides a more secure, efficient, and localized method for establishing trust channels. By employing credential injection and local authentication mechanisms, it ensures that only authorized microservices can communicate with the local security proxy service, thereby enhancing the reliability of microservice authentication. Furthermore, utilizing Unix domain sockets as the local inter-process communication mechanism significantly reduces communication latency and resource consumption. Simultaneously, leveraging operating system-level security features provides robust local security for data transmission between microservices and the local security proxy service, effectively isolating internal communication from external network risks and improving overall security and performance.

[0034] In some embodiments, the step of the local security proxy service managing the session key for a long-term secure connection includes: Store the session key for long-term secure connections in an isolated and protected memory area; When operating the session key of a long-term secure connection, anti-side-channel attack measures are adopted for protection. These measures include constant-time algorithms or algorithms that introduce randomized memory access modes to protect the session key.

[0035] Specifically, the isolated and protected memory region refers to one or more dedicated areas within the physical server's memory. These regions are configured with strict access control policies, allowing only local security agent services to perform read and write operations, and are logically or physically isolated from the memory spaces of other microservices or system processes. This can be achieved using hardware-provided memory protection mechanisms (such as Memory Management Unit (MMU) or secure enclave technology) or operating system-provided memory isolation technologies (such as process sandboxes or virtualization technology). The purpose is to prevent unauthorized access and the theft or tampering of session keys by malicious software.

[0036] The anti-side-channel attack measures can be understood as designed to defend against attacks that attempt to infer sensitive information (such as session keys) by observing system side-channel information (such as power consumption, electromagnetic radiation, time delay, cache hit rate, etc.). Specifically, constant-time algorithms ensure that the execution time remains constant or fluctuates within a very small range regardless of the input data or key value, thus preventing attackers from obtaining key information by measuring time differences. For example, during key comparison or encryption operations, even if the comparison results differ, the execution path and time length are ensured to be consistent. Randomized memory access patterns involve introducing random memory address access order, random memory access delay, or random data padding when accessing the session key in memory, making it difficult for attackers to infer key information by analyzing memory access patterns. For example, when reading or writing key data, some irrelevant memory locations can be randomly accessed, or random "decoy" data can be padded around the key data to confuse the attacker's analysis.

[0037] The technical solution of this application ensures the security of the key at the storage level by storing the session key for long-term secure connections in an isolated and protected memory area, protecting it from direct access and theft by other processes or malicious software. Furthermore, when operating the session key, anti-side-channel attack measures are employed, such as constant-time algorithms or algorithms that introduce randomized memory access patterns, effectively eliminating the possibility of attackers inferring the key by analyzing system side-channel information. It is precisely because the key is rigorously protected at both the storage and operation stages that the session key for long-term secure connections can withstand more complex attack methods, thereby significantly improving the security of the entire communication protocol.

[0038] By employing the aforementioned technical solution, this application effectively addresses the side-channel attack risk inherent in traditional session key management, significantly enhancing the security of session keys for long-term secure connections. Specifically, the isolated and protected memory region ensures key storage isolation, preventing unauthorized access; while anti-side-channel attack measures mitigate the threat of key leakage via side-channel information at the operational level. This dual protection mechanism provides a high level of security for session keys throughout their entire lifecycle, thereby improving the overall security and reliability of inter-cloud service communication and reducing the risk of security incidents due to key leakage.

[0039] In some embodiments, the step of distributing short-lived sub-session keys, wherein the sub-session keys are used for data encryption between the local security proxy service and the microservice, includes: Monitor the communication patterns of the microservices to identify whether the microservices are entering a period of sudden activity. When the microservice is detected to be entering a period of sudden activity, the asynchronous key derivation pool is started to derive the sub-session keys in batches. The sub-session keys are stored in an anti-side-channel attack memory pool and organized into multiple queues; When the microservice requests the sub-session key, the sub-session key is retrieved from the corresponding queue and distributed to the microservice through a shared memory mechanism; When receiving business data encrypted by the microservice using a sub-session key, the validity of the sub-session key is verified, and the single-use principle of the sub-session key is enforced.

[0040] Specifically, monitoring the communication patterns of microservices aims to predict or identify when they are about to enter or have already entered a high-load state, i.e., a sudden surge in activity, by analyzing the historical behavior data and real-time traffic characteristics of the microservices. For example, metrics such as the number of requests, response time, and CPU utilization of the microservices can be continuously collected, and machine learning models can be used for pattern recognition. The monitoring of the communication patterns of the microservices can be comprehensively judged by combining the health status of the microservices, system alarm information, and traffic characteristics.

[0041] When a sudden surge in activity is detected in the microservice, an asynchronous key derivation pool is activated to prevent key distribution from becoming a performance bottleneck. This pool operates independently of the main business logic and can pre-generate a large number of sub-session keys to cope with the upcoming high demand. Batch derivation can be implemented using parallel computing or multi-threading techniques to improve key generation efficiency.

[0042] To enhance the security of sub-session keys, these keys are stored in a side-channel attack resistant memory pool. This memory pool is designed to resist various side-channel attacks, such as time-based attacks and power analysis attacks. For example, techniques such as memory encryption, address randomization, and constant-time access patterns can be used to protect the keys stored therein. These keys can be organized into multiple queues in the memory pool, for example, categorized by key type, validity period, or priority of the target microservice, for efficient management and fast retrieval.

[0043] When the microservice needs to encrypt data and requests a sub-session key, the local security proxy service retrieves a pre-derived sub-session key from the corresponding queue. Subsequently, the sub-session key is distributed to the microservice via a shared memory mechanism. Due to its efficient inter-process communication capabilities, the shared memory mechanism can significantly reduce key transmission latency, making it particularly suitable for communication between the local security proxy service and microservices on the same physical server.

[0044] When receiving business data encrypted by the microservice using a sub-session key, the local security proxy service first verifies the validity of the sub-session key, such as checking if it has expired or been used before. Simultaneously, it strictly adheres to the single-use principle of the sub-session key, meaning each sub-session key is used only once to encrypt business data. Once used, the key is marked as invalid or immediately destroyed to prevent risks from replay attacks or key leakage.

[0045] Through the above technical solutions, this application significantly improves the optimization effect of secure communication protocols between cloud computing services. Specifically, by intelligently monitoring microservice communication modes and introducing asynchronous key derivation pools, the problem of low key distribution efficiency during periods of sudden microservice activity is effectively solved, key acquisition latency is greatly reduced, and smooth transmission of business data during peak periods is guaranteed. Simultaneously, the use of an anti-side-channel attack memory pool greatly enhances the storage security of sub-session keys in memory, effectively resisting potential side-channel attacks. Furthermore, the efficient distribution via shared memory mechanisms and the strict enforcement of sub-session key validity verification and single-use principles not only improve key distribution efficiency but also ensure key non-reusability from a mechanism perspective, effectively preventing replay attacks, thereby comprehensively improving the overall security and performance of communication between cloud computing services.

[0046] In some embodiments, the step of monitoring the communication pattern of the microservice and identifying whether the microservice has entered a period of sudden activity includes: Collect communication behavior data of the microservices and construct a historical behavior baseline for the microservices; The current request timestamp and request count of the microservice are obtained in real time, and real-time behavior pattern analysis is performed to obtain the real-time behavior pattern analysis results. Based on the historical behavior baseline, the threshold for judging the microservice entering a sudden period of activity is dynamically calculated and adjusted. By combining the health status of the microservices, system alarm information, and traffic characteristics, normal bursts of activity and abnormal traffic are distinguished to obtain the distinction results; By combining the real-time behavior pattern analysis results, the dynamically adjusted judgment threshold, the differentiation results, and the business priority of the microservice configuration, it is possible to identify whether the microservice has entered a period of sudden activity.

[0047] Specifically, the local security proxy service continuously records key metrics of the microservice, such as request frequency, request volume, response time, and error rate. It then performs statistical analysis and pattern recognition on this data to establish a reference model reflecting the communication characteristics of the microservice under normal operating conditions, thus constructing a historical behavior baseline for the microservice. This historical behavior baseline can be constructed using time series analysis, machine learning models (such as clustering algorithms or anomaly detection algorithms), etc., with the aim of providing a reliable reference standard for subsequent real-time behavior pattern analysis.

[0048] Specifically, the process involves acquiring the current request timestamps and request counts of the microservices in real time, performing real-time behavioral pattern analysis, and obtaining the real-time behavioral pattern analysis results. This can be understood as the local security proxy service continuously monitoring the current communication activities of the microservices, such as recording the number of received requests and the time when the requests occurred within each time window. This real-time data is then used to compare with historical behavioral baselines, analyzing the degree of deviation between the current behavioral patterns and historical patterns using statistical methods (such as moving averages and exponential smoothing) or machine learning models (such as predictive models), thereby obtaining the real-time behavioral pattern analysis results.

[0049] In practical applications, the criteria for determining whether a microservice has entered a period of sudden activity are adaptively adjusted based on the microservice's historical communication patterns and changes in the current operating environment. This allows for dynamic calculation and adjustment of the threshold for determining whether a microservice has entered a period of sudden activity. For example, if historical data shows that the microservice's traffic is inherently high during a specific time period (such as daily peak hours), the threshold will be raised accordingly to avoid misjudgment; conversely, it may be lowered during off-peak periods. This dynamic adjustment can be based on statistical methods (such as standard deviation, percentiles) or machine learning algorithms (such as reinforcement learning), with the aim of improving the accuracy and adaptability of the judgment.

[0050] Furthermore, when a traffic surge is detected, the system not only relies on the traffic data itself, but also comprehensively considers the microservice's own operational status (such as CPU utilization, memory usage, and error logs), whether there are related alarms at the system level (such as resource exhaustion alarms and network congestion alarms), and the characteristics of the traffic itself (such as request source and request type distribution). Through cross-validation of multi-dimensional information, it is possible to effectively distinguish between sudden periods of activity caused by normal business growth and abnormal traffic caused by attacks (such as DDoS attacks) or system failures. The purpose is to avoid erroneously triggering the batch distribution of sub-session keys due to abnormal traffic, or failing to distribute keys in a timely manner due to normal surges.

[0051] Therefore, all the above analysis results are weighted and fused. For example, if real-time behavior pattern analysis shows that traffic significantly exceeds the dynamically adjusted threshold, and the distinction result indicates that this is normal business growth rather than abnormal traffic, and the microservice is configured with high business priority, then it is more likely to be judged that it has entered a period of sudden activity, thus identifying whether the microservice has entered a period of sudden activity. This comprehensive judgment mechanism can provide a comprehensive, accurate, and business-sensitive identification result, ensuring that the key distribution strategy matches the actual needs and business importance of the microservice.

[0052] Through the above technical solutions, this application can significantly improve the accuracy and adaptability of identifying sudden surges in microservice activity. Specifically, the dynamically adjusted judgment threshold enables flexible responses to seasonal changes or sudden increases in microservice communication patterns, reducing false positives and false negatives. Comprehensive analysis of multi-dimensional information, particularly considering health status, system alarms, and traffic characteristics, effectively distinguishes normal business peaks from potential abnormal traffic or attacks, thereby avoiding unnecessary key distribution or security risks caused by misjudgments. Furthermore, incorporating business priority into the identification process ensures that critical business microservices can obtain sub-session keys preferentially when needed, guaranteeing the continuity and high-performance operation of core businesses. This refined identification mechanism makes the distribution of sub-session keys more accurate and efficient, thereby optimizing the overall performance and security of secure communication protocols between cloud computing services, reducing resource consumption, and improving user experience.

[0053] In some embodiments, the step of storing the sub-session key in an anti-side-channel attack memory pool and organizing it into multiple queues includes: When initializing the anti-side-channel attack memory pool, identify and select a physical memory region with an independent memory controller or low contention risk as a dedicated key storage region for the anti-side-channel attack memory pool. When writing the sub-session key into the dedicated key storage area, a fixed-size memory block is used as the storage unit for the sub-session key, and each storage unit is filled with random data. When accessing the sub-session key in the dedicated key storage area, a constant-time access mode is used; The capacity of the queue in the anti-side-channel attack memory pool is dynamically adjusted according to the business priority and real-time key requirements of the microservice. After the sub-session key is distributed, the corresponding storage unit in the key storage dedicated area is erased or reset, and random delay and memory shuffling operations are introduced.

[0054] Specifically, during the initialization of the anti-side-channel attack memory pool, physical memory regions with independent memory controllers or low contention risk are identified and selected as dedicated key storage areas for the anti-side-channel attack memory pool. Identifying physical memory regions with independent memory controllers or low contention risk aims to provide a higher level of hardware isolation and resource protection for sensitive sub-session keys. For example, in multi-core processor or multi-socket systems, certain memory regions may be managed by independent memory controllers, or their access paths may be separated from the access paths of other high-contention components, thereby significantly reducing the risk of side-channel attackers inferring key information by monitoring memory access patterns. By storing keys in such dedicated areas, memory access conflicts with other tenants or system processes can be effectively reduced, further enhancing key security.

[0055] Furthermore, when writing the sub-session key into the dedicated key storage area, fixed-size memory blocks are used as storage units for the sub-session key, and each storage unit is randomly padded with data. Using fixed-size memory blocks as storage units prevents attackers from inferring the key length or type by observing the size of memory allocation or deallocation. Random data padding involves filling the remaining portion of the memory block with random, meaningless data outside the actual key data. Its purpose is to obscure the actual boundaries and content of the key, making it difficult for a side-channel attacker to identify the key's existence or extract its information by analyzing changes in entropy or data patterns in the memory area.

[0056] Specifically, a constant-time access mode is employed when accessing the sub-session key in the dedicated key storage area. Constant-time access mode means that the time required to access the key is fixed regardless of its specific value or storage location. This mode effectively resists time-difference-based side-channel attacks, where attackers infer key bit information by measuring memory access time. By ensuring consistent time overhead for each key access, time side-channeling is eliminated even with caching or memory access optimizations, thereby protecting key confidentiality.

[0057] In practical applications, the capacity of the queues in the anti-side-channel attack memory pool is dynamically adjusted based on the business priority and real-time key requirements of the microservices. This aims to optimize resource utilization and ensure timely key supply. For example, for microservices with high business priority or those expected to enter a period of sudden activity, their corresponding key queue capacity can be increased in advance to cope with high-concurrency key requests and avoid performance degradation due to key shortages. This dynamic adjustment mechanism can flexibly allocate memory resources according to actual operating conditions, improving responsiveness and overall efficiency.

[0058] Furthermore, after distributing the sub-session key, the corresponding storage unit in the dedicated key storage area is erased or reset, and random delays and memory shuffling operations are introduced. Key erasure or reset ensures that any residual traces of the key in memory are completely removed after it is no longer used, preventing attackers from recovering used keys through memory forensics techniques. Introducing random delays and memory shuffling further obfuscates the timing and location of the erasure operation, making it difficult for attackers to identify the timing and scope of key erasure by observing memory bus activity or cache status. This enhances the security of key destruction and effectively combats advanced memory forensics and side-channel attacks.

[0059] The technical solution of this application significantly enhances the security and management efficiency of sub-session keys in a side-channel attack-resistant memory pool through a series of refined memory management and access strategies. First, by identifying and selecting physical memory regions with independent memory controllers or low contention risk as dedicated key storage areas, hardware-level isolation of the keys is provided, effectively reducing memory access conflicts with other processes or tenants. This reduces the possibility of side-channel attackers inferring key information by observing shared resource contention. Second, using fixed-size memory blocks to store sub-session keys and randomly padding them with data obscures the actual length and content of the keys, making it difficult for attackers to identify the keys through memory layout or data pattern analysis. Simultaneously, the introduction of a constant-time access mode eliminates time-difference-based side-channel attack vectors, ensuring the indistinguishability of key access operations. Furthermore, dynamically adjusting the queue capacity based on the microservice's business priority and real-time key requirements not only optimizes memory resource allocation and ensures timely key supply but also avoids resource waste or performance bottlenecks that might result from fixed capacity. Finally, the key is erased or reset immediately after distribution, and combined with random delay and memory shuffling operations, to ensure the complete destruction of the key at the end of its life cycle. This effectively combats memory forensics attacks and further obfuscates the destruction process by introducing randomness, making it difficult for attackers to recover the used key by analyzing memory operation traces.

[0060] Through the above technical solutions, this application can significantly improve the storage security of sub-session keys in inter-cloud computing service communication, effectively resisting various side-channel attacks and memory forensics attacks. Specifically, by using hardware-level memory isolation and low-contention region selection, the risk of key leakage is reduced; by using fixed-size storage and random data padding, the anti-analysis capability of key data is enhanced; by using constant-time access mode, the threat of time-side channel attacks is eliminated; by using dynamic queue capacity adjustment, key distribution efficiency and resource utilization are optimized; and by introducing a key erasure mechanism with random latency and memory shuffling, the complete destruction of keys is ensured, further improving the security of key lifecycle management. These improvements collectively construct a more robust and secure key storage and management system, which is particularly suitable for cloud computing environments with extremely high security requirements, effectively solving the key storage security risks and efficiency problems that may exist in basic technical solutions.

[0061] In some alternative embodiments, it is assumed that a local security proxy service is deployed on a physical server to provide secure communication for multiple microservices. When one of the microservices, such as a high-concurrency e-commerce order processing service, is identified as entering a period of sudden activity, the local security proxy service will start an asynchronous key derivation pool and generate a large number of short-lifetime sub-session keys in batches.

[0062] Specifically, when initializing the anti-side-channel attack memory pool, the local security agent service identifies and requests the allocation of a physical memory region under an independent memory controller through the operating system or hardware abstraction layer interface, or selects a memory region with the lowest current system contention risk as a dedicated key storage region for storing these sub-session keys. For example, in a NUMA architecture, a dedicated region on the memory node with the highest affinity to the local security agent service's CPU core might be selected.

[0063] When a derived sub-session key needs to be written to this dedicated area, each sub-session key is encapsulated in a predefined, fixed-size memory block, such as 64 bytes. If the key itself is smaller than 64 bytes, the remaining space is filled with cryptographically secure random numbers to ensure that all storage units are of consistent size and highly random in content, thereby preventing attackers from inferring key information by observing memory block size or content patterns.

[0064] When a microservice requests a sub-session key, the local security proxy service retrieves the key from the corresponding queue and distributes it via a shared memory mechanism. When accessing these keys, regardless of their specific location in memory or their value, the access operation is designed to have a constant time consumption, for example, by forcibly reading a fixed number of memory addresses or executing a fixed number of CPU instructions, to eliminate any potential time side-channeling.

[0065] Once the sub-session key is distributed and the microservice confirms receipt, the local security proxy service immediately performs an erase operation on the corresponding memory cell in the dedicated key storage area, for example, overwriting the memory area with zeros or random data. To further enhance security, the erase operation introduces a random delay, such as a delay time randomly selected between 50 and 150 milliseconds. Furthermore, a memory shuffling operation may be triggered before and after the erase, i.e., remapping or moving memory pages through the Memory Management Unit (MMU), making the physical location of the key unpredictable. This significantly increases the difficulty for attackers to recover used keys using memory forensics techniques. Through these measures, even in extreme cases, attackers find it difficult to obtain a valid sub-session key from memory.

[0066] In some embodiments described above in this application, after distributing the sub-session key, the corresponding storage unit in the key storage dedicated area is erased or reset, and random latency and memory shuffling operations are introduced. While this effectively prevents side-channel attacks, in a dynamically changing cloud computing environment, factors such as the real-time contention level of physical memory resources, system load, and memory controller queue depth will continuously fluctuate. If the execution mode, random latency time range, and distribution of the key erasure operation are fixed, it may not be able to adequately adapt to these dynamic changes. This can lead to reduced efficiency of the erasure operation in certain high-contention or high-load scenarios, weakened resistance to side-channel attacks, or even failure to detect anomalies or potential attack attempts in the erasure operation in a timely manner due to the lack of a real-time verification mechanism.

[0067] In some embodiments, the step of erasing or resetting the storage unit corresponding to the key storage dedicated area after distributing the sub-session key, and introducing random latency and memory shuffling operations, includes: The execution mode of the erase or reset operation is dynamically adjusted based on the real-time contention level of physical memory resources and the business priority of the microservice. The time range and distribution of random latency are dynamically adjusted based on system load and memory controller queue depth. Combining the memory page remapping function provided by the memory management unit, the physical memory pages of the key storage dedicated area are remapped in address and moved in physical location. Monitor memory access latency, cache hit rate, and memory bandwidth usage; detect deviations between the execution patterns of erase, reset, random latency, or memory shuffle operations and the expected safe behavior patterns; trigger alerts and adjust operating parameters or isolate memory regions.

[0068] The local security proxy service obtains real-time information on the memory resource usage on the physical server, the frequency and intensity of memory access by different microservices, and intelligently selects the specific strategy for erasure or reset operations based on the business importance of the microservice being processed (e.g., high-priority services may require faster and more thorough erasure, while low-priority services can be appropriately relaxed while ensuring security). This allows for dynamic adjustment of the execution mode of erasure or reset operations. For example, when memory contention is intense, a lighter-weight reset operation may be chosen to reduce the impact on performance, while when resources are sufficient or highly sensitive keys are involved, multi-round overwrite erasure may be used to ensure complete data destruction.

[0069] Furthermore, the phrase "dynamically adjusting the time range and distribution of random delay based on system load and memory controller queue depth" can be understood as follows: to further obfuscate the analysis of key erasure timing by side-channel attackers, the local security proxy service introduces a random delay after performing an erasure or reset operation. The length of this delay and its distribution pattern within a certain range are not fixed but are adjusted in real-time based on the overall system load of the physical server (such as CPU utilization and I / O throughput) and the request queue depth within the memory controller. For example, when the system load is high or the memory controller queue is deep, the average value of the random delay can be appropriately extended or its distribution range expanded to better integrate with system noise, making it difficult for attackers to distinguish between normal system delays and deliberately introduced security delays.

[0070] In practical applications, the Memory Management Unit (MMU) function provided by the operating system or hardware virtualization layer is utilized to dynamically change the mapping relationship between the logical addresses and physical addresses of the physical memory pages storing sub-session keys. This can even involve migrating these physical memory pages from one physical location to another, achieving address remapping and physical location migration of the physical memory pages in the dedicated key storage area. The purpose is to ensure that even if an attacker locks a logical address, they cannot determine its corresponding physical location, thus effectively combating side-channel attacks based on physical addresses or cache locations, and further enhancing the isolation and security of the key storage area.

[0071] In addition, the local security agent service continuously collects performance metrics related to memory operations, such as memory access request response time, CPU cache hit rate, and memory bus bandwidth usage. By comparing this real-time monitoring data with preset, security-expected behavior patterns, any anomalies can be detected promptly, triggering alerts and adjusting operating parameters or isolating memory regions. For example, if memory access latency abnormally increases after an erase operation, or if the cache hit rate drops unexpectedly, this may indicate that the erase operation failed to perform as expected, or that a potential side-channel attack attempt exists. Once a significant deviation is detected, a security alert will be immediately triggered, and the parameters for erase, reset, random delay, or memory shuffling operations will be automatically adjusted according to the severity of the deviation to restore a secure state, or, in extreme cases, the abnormal memory region will be directly isolated to prevent key information leakage.

[0072] This application's technical solution effectively addresses the static nature, predictability, and lack of real-time verification issues inherent in basic technical solutions by introducing a dynamic adaptive key erasure mechanism, memory page remapping technology, and real-time security monitoring and response capabilities. Specifically, while basic technical solutions propose erasure / reset and introduce random delays, if these operations are fixed, attackers could potentially find patterns through long-term observation and statistical analysis, thereby deducing key information. This application dynamically adjusts the execution mode of erasure or reset operations based on the real-time contention level of physical memory resources and the business priority of the microservice. This allows the strength and method of key erasure to intelligently adapt to the actual operating environment and business needs, avoiding the predictability risks associated with fixed patterns and optimizing the balance between performance and security.

[0073] Furthermore, the time range and distribution of random latency are dynamically adjusted based on system load and memory controller queue depth, ensuring that the introduction of random latency is more covert and effective, allowing it to better blend into inherent noise and significantly increasing the difficulty for attackers to infer key information through timing analysis. In addition, combined with the memory page remapping function provided by the memory management unit, the physical memory pages of the dedicated key storage area are remapped in address and their physical locations are migrated, fundamentally changing the physical location of key storage. Even if an attacker can lock the logical address through some means, they cannot continuously track its location in physical memory, thus effectively combating side-channel attacks based on physical addresses or cache locations.

[0074] Most importantly, by monitoring memory access latency, cache hit rate, and memory bandwidth usage, and detecting deviations between the execution patterns of erase, reset, random latency, or memory shuffling operations and expected security behavior patterns, a proactive security verification and response mechanism is provided. This allows the local security agent service to move beyond passively performing cleanup operations and instead assess their effectiveness in real time, promptly detecting and responding to any potential anomalies, whether due to operational failures or external attack attempts. It is this combination of dynamic adjustment, physical isolation, and real-time monitoring that makes the post-distribution cleanup process of sub-session keys more robust, unpredictable, and self-protective, significantly enhancing the security of the entire cloud computing service inter-service secure communication protocol.

[0075] Through the above technical solutions, this application can significantly improve the security and robustness of sub-session key erasure in secure communication between cloud computing services. First, dynamically adjusting the execution mode of erase or reset operations allows the key erasure strategy to be optimized based on real-time resource contention and business priorities, avoiding performance bottlenecks or security vulnerabilities that may arise from fixed strategies and ensuring the effectiveness of key erasure under different operating environments. Second, dynamically adjusting the time range and distribution of random delays greatly increases the difficulty for side-channel attackers to infer key information through timing analysis, effectively enhancing the ability to resist timing attacks. Furthermore, combined with the memory page remapping function of the memory management unit, the physical location of the key storage area is dynamically migrated, fundamentally combating side-channel attacks based on physical addresses or cache locations, and providing a deeper level of isolation protection for the keys. Most importantly, by continuously monitoring memory access latency, cache hit rate, and memory bandwidth usage, and detecting deviations between operation execution patterns and expected security behavior patterns, this application can detect and respond to potential security threats or operational anomalies in real time, triggering alarms, adjusting parameters, or isolating memory regions in a timely manner. This constructs a proactive and adaptive key eviction mechanism, significantly reducing the risk of key leakage and improving overall security and reliability.

[0076] In some embodiments, the steps of monitoring memory access latency, cache hit rate, and memory bandwidth usage, detecting deviations between the execution patterns of erase, reset, random latency, or memory shuffling operations and the expected safe behavior patterns, triggering alarms, and adjusting operating parameters or isolating memory regions include: Continuously collect monitoring metrics data such as memory access latency, cache hit rate, and memory bandwidth usage, and perform multi-dimensional baseline modeling on the collected raw data to obtain a multi-dimensional baseline. The multi-dimensional baseline modeling takes into account the historical behavior patterns of the microservice, system load, and real-time contention status of physical memory resources. The monitoring indicator data is acquired in real time and compared with a multi-dimensional baseline to calculate the deviation between the monitoring indicator data and the multi-dimensional baseline. Based on the magnitude and duration of the deviation, combined with the business priority and system alarm information of the microservice configuration, it is determined whether the deviation is caused by an abnormal execution mode of erase, reset, random delay or memory shuffle operation, or by interference from other tenants or system components. When it is determined that the deviation is caused by an abnormal execution mode, the local security agent service triggers an alarm and adjusts the operating parameters or isolates the memory region according to the severity of the deviation. When it is determined that the deviation is caused by interference from other tenants or system components, the local security agent service records the interference event but does not trigger a security alert, and adjusts the baseline model to adapt to the new interference pattern, or reports the interference to the cloud platform resource scheduling system.

[0077] Specifically, the local security agent service periodically collects various memory access-related metrics through the operating system's performance monitoring interface or hardware performance counters, continuously gathering monitoring data on memory access latency, cache hit rate, and memory bandwidth usage. This raw data forms the basis for subsequent analysis, and its collection frequency and granularity can be dynamically adjusted according to system load and security requirements.

[0078] The process involves multi-dimensional baseline modeling of the collected raw data to obtain a multi-dimensional baseline. This multi-dimensional baseline modeling considers the historical behavior patterns of the microservices, system load, and real-time contention status of physical memory resources. This means that the baseline model is not a single static value, but a dynamic model that integrates multiple influencing factors. For example, the historical behavior patterns of the microservices can include their typical memory access characteristics in different time periods (such as weekdays, weekends, peak periods, and off-peak periods); system load can refer to CPU utilization, network I / O, etc.; and the real-time contention status of physical memory resources reflects the competition for memory by other tenants or processes in a shared memory environment. Machine learning algorithms (such as time series analysis and cluster analysis) can be used to construct a multi-dimensional baseline that reflects the complex memory behavior under normal operating conditions.

[0079] In practical applications, the monitoring indicator data is acquired in real time and compared with a multi-dimensional baseline to calculate the deviation between the monitoring indicator data and the multi-dimensional baseline. This step aims to quantify the difference between current memory behavior and expected normal behavior. The deviation can be calculated using various statistical methods, such as standard deviation, percentage difference, or more complex anomaly detection algorithms.

[0080] Furthermore, based on the magnitude and duration of the deviation, combined with the business priority configured in the microservice and system alarm information, it is determined whether the deviation is caused by an abnormal execution mode of erase, reset, random delay, or memory shuffle operations, or by interference from other tenants or system components. This determination process is the core of this technical solution. A large deviation magnitude and long duration may be more indicative of an abnormal execution mode, while short-term fluctuations may be caused by interference. Microservices with high business priorities are more sensitive to performance, and their deviations may require more stringent judgment. System alarm information (such as resource overrun alarms from other tenants, underlying hardware failure alarms) can serve as auxiliary judgment criteria.

[0081] When the deviation is determined to be caused by an abnormal execution mode, the local security agent service triggers an alert and adjusts operating parameters or isolates memory regions based on the severity of the deviation. This means that proactive security response measures will be taken; for example, if an erase operation is detected not being executed at a constant time, the erase algorithm parameters may be adjusted, or in severe cases, the affected memory region may be isolated to prevent sensitive data leakage.

[0082] Conversely, when the deviation is determined to be caused by interference from other tenants or system components, the local security agent service records the interference event without triggering a security alert, and adjusts the baseline model to adapt to the new interference pattern, or reports the interference to the cloud platform resource scheduling system. This differentiated response avoids unnecessary security alerts and resource isolation, improving stability and resource utilization. Recording interference events aids in subsequent analysis and optimization, adjusting the baseline model enables learning and adaptation to new normal operating modes, and reporting to the cloud platform helps the resource scheduling system to allocate resources more rationally and resolve conflicts.

[0083] This application significantly improves the security and robustness of the cloud computing service inter-service secure communication protocol optimization method in complex multi-tenant environments. First, multi-dimensional baseline modeling and real-time comparison mechanisms enable more accurate detection of memory access pattern anomalies, reducing false alarm rates. Second, intelligent judgment combining deviation characteristics, business priorities, and system alarm information accurately identifies the root cause of deviations, avoiding misjudging external interference as internal security anomalies, thereby reducing unnecessary security alerts and resource isolation operations. Third, differentiated response strategies are adopted for different causes: triggering alerts and adjusting parameters or isolating resources for execution pattern anomalies, while recording events, adjusting baselines, or reporting to the resource scheduling system for external interference. This makes security responses more accurate and efficient, effectively addressing potential security threats while maintaining stable operation and resource utilization, thus improving overall security protection capabilities and operational efficiency.

[0084] In some embodiments, the step of determining whether the deviation is caused by an abnormal execution mode of erase, reset, random delay, or memory shuffle operations, or by interference from other tenants or system components, based on the magnitude and duration of the deviation and in conjunction with the business priority and system alarm information of the microservice configuration, includes: The deviation is matched with a variety of preset deviation patterns to identify the deviation pattern to which the current deviation belongs; Based on the identified deviation patterns, a set of predefined judgment rules is selected. The judgment rule set defines the weight factors and confidence thresholds for business priorities and system alarm information for different deviation patterns. Based on the selected set of judgment rules, the business priority and the system alarm information are weighted and analyzed, and the weighting factors and the confidence threshold are dynamically adjusted in combination with historical misjudgments and missed reports. Based on the results of the weighted analysis, the probability that the deviation is caused by abnormal execution mode and the probability that it is caused by interference from other tenants or system components are calculated. The calculated probability is compared with the dynamically adjusted confidence threshold to obtain the result of the deviation source judgment; If the determination of the source of the deviation is still ambiguous, the local security agent service will initiate a cross-agent collaborative verification mechanism to request relevant monitoring data and judgment results from other local security agent services on the same physical server or in the cluster for cross-verification.

[0085] Specifically, the deviations in real-time detected monitoring metrics such as memory access latency, cache hit rate, and memory bandwidth usage are compared with predefined patterns representing different types of anomalies or interference. This allows for matching the deviations with multiple preset deviation patterns. These patterns can be constructed based on historical data analysis, expert experience, or machine learning models; for example, they may include "persistent high latency patterns," "periodic cache jitter patterns," or "burst bandwidth usage patterns." Pattern matching allows for a more refined identification of the nature of the current deviation. The preset deviation patterns can be understood as a series of feature vectors or time-series templates used to describe specific types of memory behavior anomalies or external interference. For example, one deviation pattern might describe a periodic, small fluctuation in memory access latency within a specific time window, along with a slight decrease in cache hit rate, which could be related to a specific side-channel attack attempt.

[0086] Based on the identified deviation patterns, a predefined set of judgment rules is selected. This set defines weighting factors and confidence thresholds for business priorities and system alarm information for different deviation patterns. This means that different judgment logic and parameters will be used for different deviation patterns. For example, for a deviation identified as a "persistent high latency pattern," the judgment rule set may assign higher weight to business priorities because persistent high latency may have a greater impact on critical business operations; while for a "periodic cache jitter pattern," more emphasis may be placed on system alarm information to determine whether there are anomalies in the underlying hardware or virtualization layer. The weighting factors are used to quantify the relative importance of business priorities and system alarm information in the judgment process, while the confidence threshold is used to determine the reliability standard of the judgment result.

[0087] Furthermore, based on the selected set of judgment rules, a weighted analysis is performed on the business priority and the system alarm information. Combined with historical misjudgments and missed detections, the weighting factors and the confidence threshold are dynamically adjusted. Weighted analysis is a comprehensive evaluation method that accumulates or fuses various pieces of information by multiplying them by their corresponding weighting factors to obtain a comprehensive judgment score. For example, different scores can be assigned based on the level of business priority, and different scores can be assigned based on the severity of system alarms. These scores are then summed with weights. Combining historical misjudgments and missed detections creates a feedback loop. For example, if a certain deviation pattern is frequently misjudged as an abnormal execution mode, the weight of the business priority in that pattern can be appropriately reduced, or the confidence threshold can be adjusted to improve the accuracy of the judgment. This dynamic adjustment mechanism allows the judgment system to continuously learn and optimize.

[0088] Based on this, the probability that the deviation is caused by an abnormal execution mode and the probability that it is caused by interference from other tenants or system components are calculated according to the results of the weighted analysis. This can be achieved through Bayesian inference, logistic regression, or other probabilistic models, taking the results of the weighted analysis as input and outputting probability values ​​for the two possible causes. For example, if the weighted analysis results strongly point to an abnormal execution mode, the calculated probability of an abnormal execution mode will be significantly higher than the probability of interference.

[0089] The calculated probabilities are then compared with the dynamically adjusted confidence thresholds to determine the source of the deviation. If the probability of an abnormal execution mode is higher than its corresponding confidence threshold, it is determined to be an abnormal execution mode; conversely, if the probability of interference is higher than its corresponding confidence threshold, it is determined to be caused by interference. If both probabilities are lower than their respective thresholds, or if the two probabilities are very close, there may be ambiguity in the judgment.

[0090] If the determination of the source of the deviation remains ambiguous, the local security proxy service initiates a cross-proxy collaborative verification mechanism. This mechanism requests relevant monitoring data and judgment results from other local security proxy services on the same physical server or within the cluster for cross-verification. The cross-proxy collaborative verification mechanism is a distributed decision enhancement method. When a single local security proxy service cannot clearly determine the source of the deviation, it can request data and judgment results from other local security proxy services with similar monitoring capabilities or in a relevant context. For example, if another local security proxy service on the same physical server also reports a similar memory access anomaly, or if related microservices on other servers in the cluster also exhibit anomalies, this information can serve as supplementary evidence to help eliminate ambiguity in the judgment and improve the accuracy and reliability of the final determination.

[0091] The technical solution of this application introduces deviation pattern matching, dynamically adjusted judgment rule sets, weighted analysis, and probability calculation, making the judgment of the source of memory behavior deviations more refined and intelligent. When a single local security agent service cannot reach a clear conclusion, a cross-agent collaborative verification mechanism can utilize multi-source information in a distributed environment for cross-comparison, thereby effectively solving the ambiguity problem in judging the source of deviations in complex cloud environments. This multi-layered, adaptive judgment mechanism significantly improves the ability to identify potential side-channel attacks or system interference.

[0092] The above technical solutions significantly improve the accuracy and robustness of identifying the sources of memory operation anomalies. Through refined deviation pattern matching and dynamically adjusted judgment rule sets, different types of anomalies and interferences can be effectively distinguished, reducing false positives and false negatives. Weighted analysis and probability calculation provide more quantitative judgment criteria, making the decision-making process more scientific. Especially when the judgment results are ambiguous, the cross-agent collaborative verification mechanism can introduce broader contextual information, eliminating uncertainty through multi-party consensus or cross-validation, thereby avoiding resource waste (such as unnecessary isolation) or security risks (such as failure to detect attacks in a timely manner) caused by misjudgment. This makes the optimized method for secure communication protocols between cloud computing services more adaptable and protective against complex and ever-changing threats.

[0093] This application also proposes an optimization system for secure communication protocols between cloud computing services, such as... Figure 2 As shown, a cloud computing service inter-service secure communication protocol optimization system 100 includes: Deployment module 10 is used to deploy a local security proxy service on the physical server hosting the microservice and configure resource isolation for the local security proxy service to ensure that its operation stability and performance are not affected by the resource contention of the microservice on the physical server; Trust channel establishment module 20 establishes a trust channel between the local security proxy service and the microservice, the trust channel being used for authentication and data transmission of the microservice; The handshake and key management module 30 is used for the local security proxy service to perform a secure communication protocol handshake with the external service to establish a long-term secure connection, and the local security proxy service manages the session key of the long-term secure connection. The sub-session key distribution module 40 is used to distribute short-lifetime sub-session keys to the local security proxy service, and the sub-session keys are used for data encryption between the local security proxy service and the microservice. Data processing module 50 is used for the local security proxy service to receive business data encrypted by the microservice using a sub-session key through the trusted channel, the local security proxy service to decrypt the business data, re-encrypt the business data using the session key of a long-term secure connection, and forward it to the external service.

[0094] This application constructs a modular system architecture to effectively isolate and manage the complexity, resource consumption, and security risks of secure communication between cloud computing services. By deploying modules to isolate local security proxy services, it ensures that their critical security functions are not affected by microservice resource contention, thereby guaranteeing system stability and performance. Simultaneously, the trust channel establishment module, handshake and key management module, sub-session key distribution module, and data processing module work collaboratively to implement a layered key management and proxy communication mechanism. This reduces the security processing burden on microservices and effectively resists side-channel attacks through short-lifespan sub-session keys, ultimately achieving a balance between performance and security, and improving the overall security protection capabilities and operational efficiency of the cloud platform.

[0095] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for optimizing a secure communication protocol between cloud computing services, characterized in that, The method is executed by a local security proxy service deployed on the physical server hosting the microservice, and includes: Establish a trust channel with the microservice, the trust channel being used for authentication and data transmission of the microservice; The local security proxy service performs a secure communication protocol handshake with external services to establish a long-term secure connection, and manages the session key of the long-term secure connection. Distribute short-lived sub-session keys, which are used for data encryption between the local security proxy service and the microservice; The microservice receives business data encrypted with a sub-session key through the trusted channel. The local security proxy service decrypts the business data, re-encrypts it using the session key of the long-term secure connection, and forwards it to the external service.

2. The method for optimizing secure communication protocols between cloud computing services according to claim 1, characterized in that, The steps for establishing a trust channel with the microservice specifically include: Inject credentials for local authentication into the microservice; The microservice uses the credentials to initiate authentication with the local security proxy service; Once authentication is successful, a secure connection based on a local inter-process communication mechanism is established as a trusted channel between the local security proxy service and the microservice. The local inter-process communication mechanism is a Unix domain socket.

3. The method for optimizing secure communication protocols between cloud computing services according to claim 1, characterized in that, The steps of the local security proxy service managing the session keys for long-term secure connections include: Store the session key for long-term secure connections in an isolated and protected memory area; When operating the session key of a long-term secure connection, anti-side-channel attack measures are adopted for protection. These measures include constant-time algorithms or algorithms that introduce randomized memory access modes to protect the session key.

4. The method for optimizing secure communication protocols between cloud computing services according to claim 1, characterized in that, The step of distributing short-lived sub-session keys, wherein the sub-session keys are used for data encryption between the local security proxy service and the microservice, includes: Monitor the communication patterns of the microservices to identify whether the microservices are entering a period of sudden activity. When the microservice is detected to be entering a period of sudden activity, the asynchronous key derivation pool is started to derive the sub-session keys in batches. The sub-session keys are stored in an anti-side-channel attack memory pool and organized into multiple queues; When the microservice requests the sub-session key, the sub-session key is retrieved from the corresponding queue and distributed to the microservice through a shared memory mechanism; When receiving business data encrypted by the microservice using a sub-session key, the validity of the sub-session key is verified, and the single-use principle of the sub-session key is enforced.

5. The method for optimizing secure communication protocols between cloud computing services according to claim 4, characterized in that, The step of monitoring the communication pattern of the microservice and identifying whether the microservice has entered a period of sudden activity includes: Collect communication behavior data of the microservices and construct a historical behavior baseline for the microservices; The current request timestamp and request count of the microservice are obtained in real time, and real-time behavior pattern analysis is performed to obtain the real-time behavior pattern analysis results. Based on the historical behavior baseline, the threshold for judging the microservice entering a sudden period of activity is dynamically calculated and adjusted. By combining the health status of the microservices, system alarm information, and traffic characteristics, normal bursts of activity and abnormal traffic are distinguished to obtain the distinction results; By combining the real-time behavior pattern analysis results, the dynamically adjusted judgment threshold, the differentiation results, and the business priority of the microservice configuration, it is determined whether the microservice has entered a period of sudden activity.

6. The method for optimizing secure communication protocols between cloud computing services according to claim 4, characterized in that, The step of storing the sub-session key in an anti-side-channel attack memory pool and organizing it into multiple queues includes: When initializing the anti-side-channel attack memory pool, identify and select a physical memory region with an independent memory controller or low contention risk as a dedicated key storage region for the anti-side-channel attack memory pool. When writing the sub-session key into the dedicated key storage area, a fixed-size memory block is used as the storage unit for the sub-session key, and each storage unit is filled with random data. When accessing the sub-session key in the dedicated key storage area, a constant-time access mode is used; The capacity of the queues in the anti-side-channel attack memory pool is dynamically adjusted according to the business priority and real-time key requirements of the microservices. After the sub-session key is distributed, the corresponding storage unit in the key storage dedicated area is erased or reset, and random delay and memory shuffling operations are introduced.

7. The method for optimizing secure communication protocols between cloud computing services according to claim 6, characterized in that, The steps of erasing or resetting the corresponding storage unit in the dedicated key storage area after distributing the sub-session key, and introducing random delay and memory shuffling operations, include: The execution mode of the erase or reset operation is dynamically adjusted based on the real-time contention level of physical memory resources and the business priority of the microservice. The time range and distribution of random latency are dynamically adjusted based on system load and memory controller queue depth. Combining the memory page remapping function provided by the memory management unit, the physical memory pages of the key storage dedicated area are remapped in address and moved in physical location. Monitor memory access latency, cache hit rate, and memory bandwidth usage; detect deviations between the execution patterns of erase, reset, random latency, or memory shuffle operations and the expected safe behavior patterns; trigger alerts and adjust operating parameters or isolate memory regions.

8. The method for optimizing secure communication protocols between cloud computing services according to claim 7, characterized in that, The steps of monitoring memory access latency, cache hit rate, and memory bandwidth usage, detecting deviations between the execution mode of erase, reset, random latency, or memory shuffling operations and the expected safe behavior mode, and triggering alarms and adjusting operating parameters or isolating memory regions include: Continuously collect monitoring metrics data such as memory access latency, cache hit rate, and memory bandwidth usage, and perform multi-dimensional baseline modeling on the collected raw data to obtain a multi-dimensional baseline. The multi-dimensional baseline modeling takes into account the historical behavior patterns of the microservice, system load, and real-time contention status of physical memory resources. The monitoring indicator data is acquired in real time and compared with a multi-dimensional baseline to calculate the deviation between the monitoring indicator data and the multi-dimensional baseline. Based on the magnitude and duration of the deviation, combined with the business priority and system alarm information of the microservice configuration, it is determined whether the deviation is caused by an abnormal execution mode of erase, reset, random delay or memory shuffle operation, or by interference from other tenants or system components. When it is determined that the deviation is caused by an abnormal execution mode, the local security agent service triggers an alarm and adjusts the operating parameters or isolates the memory region according to the severity of the deviation. When it is determined that the deviation is caused by interference from other tenants or system components, the local security agent service records the interference event but does not trigger a security alert, and adjusts the baseline model to adapt to the new interference pattern, or reports the interference to the cloud platform resource scheduling system.

9. The method for optimizing secure communication protocols between cloud computing services according to claim 8, characterized in that, The step of determining whether the deviation is caused by an abnormal execution mode of erase, reset, random delay, or memory shuffle operations, or by interference from other tenants or system components, based on the magnitude and duration of the deviation and in conjunction with the business priority and system alarm information configured in the microservice, includes: The deviation is matched with a variety of preset deviation patterns to identify the deviation pattern to which the current deviation belongs; Based on the identified deviation patterns, a set of predefined judgment rules is selected. The judgment rule set defines the weight factors and confidence thresholds for business priorities and system alarm information for different deviation patterns. Based on the selected set of judgment rules, the business priority and the system alarm information are weighted and analyzed, and the weighting factors and the confidence threshold are dynamically adjusted in combination with historical misjudgments and missed reports. Based on the results of the weighted analysis, the probability that the deviation is caused by abnormal execution mode and the probability that it is caused by interference from other tenants or system components are calculated. The calculated probability is compared with the dynamically adjusted confidence threshold to obtain the result of the deviation source judgment; If the determination of the source of the deviation is still ambiguous, the local security agent service will initiate a cross-agent collaborative verification mechanism to request relevant monitoring data and judgment results from other local security agent services on the same physical server or in the cluster for cross-verification.

10. A system for optimizing secure communication protocols between cloud computing services, characterized in that, The system includes: The deployment module is used to deploy a local security proxy service on the physical server hosting the microservices and configure resource isolation for the local security proxy service to ensure that its operational stability and performance are not affected by resource contention of the microservices on the physical server; The trust channel establishment module establishes a trust channel between the local security proxy service and the microservice, and the trust channel is used for the authentication and data transmission of the microservice. The handshake and key management module is used for the local security proxy service to perform a secure communication protocol handshake with the external service to establish a long-term secure connection, and the local security proxy service manages the session key of the long-term secure connection. The sub-session key distribution module is used by the local security proxy service to distribute short-lifetime sub-session keys, which are used for data encryption between the local security proxy service and the microservice. The data processing module is used for the local security proxy service to receive business data encrypted by the microservice using a sub-session key through the trusted channel, the local security proxy service to decrypt the business data, re-encrypt the business data using the session key of a long-term secure connection, and forward it to the external service.

Images