A hybrid storage access scheduling method, system and medium for storage virtualization

Abstract

The application relates to the technical field of data processing, and discloses a hybrid storage access scheduling method and system for storage virtualization and a medium. The method comprises the following steps: collecting a write access request, a snapshot chain depth, a hotness flip frequency and a time interval. The request urgency is determined according to the write access request, and it is judged whether the temporary acceleration processing condition is met. When the condition is met, a temporary acceleration mapping relationship is established, and the access request is guided to the temporary acceleration mapping relationship. The access frequency of a target logical block segment, the to-be-written state of a high-speed storage layer and the snapshot chain additional write amount are collected, and it is judged whether the temporary acceleration mapping relationship meets the formal migration condition. When it is judged that the formal migration condition is met, the temporary acceleration mapping relationship is converted into a formal mapping relationship. Otherwise, the temporary acceleration mapping relationship is cancelled, and the main mapping relationship remains unchanged. According to the application, the high-urgency write access response and migration smoothness are improved while the main mapping relationship is kept stable.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and more specifically, to a hybrid storage access scheduling method, system, and medium for storage virtualization. Background Technology

[0002] With the continuous development of cloud computing, virtual machine clusters, distributed storage, and data center infrastructure, storage virtualization technology has been widely applied in scenarios such as decoupling computing and storage resources, unified management of virtual volumes, multi-layer media collaborative access, and snapshot and clone data protection. In such systems, logical volumes often need to simultaneously address the differentiated media characteristics of high-speed storage layers and capacity storage layers, while also considering the write access latency of virtual machine services, the metadata management overhead brought by snapshot chains, and the write-back pressure under continuous write conditions. Therefore, how to achieve fine-grained access scheduling for hybrid storage within the storage virtualization system has become a crucial technical issue in the field of computer storage management and input / output control.

[0003] In existing technologies, such as invention patent CN112667356A, latency predictability in virtualized environments is improved through physical queue allocation, rate limiting threshold calculation, and queue reallocation. Another example is CN110825324A, which primarily relies on data block access information to perform migration between different storage media to optimize capacity and performance configuration in hybrid storage systems. While these solutions improve virtualization queue management and hot / cold data migration, in storage virtualization environments, write access to logical block segments within snapshot chains still presents several problems: First, the lack of unified constraints between the urgency of write access requests, the additional write overhead from snapshot chains, and the timing of media layer migration leads to frequent migrations during short-term hotspot fluctuations. Second, when there is write-back pressure at high-speed storage layers, existing methods of directly processing based on access characteristics or migration conditions can easily result in increased write amplification, fluctuating access latency, and insufficient migration stability.

[0004] Therefore, it is necessary to design a hybrid storage access scheduling method, system, and medium for storage virtualization to solve the problems existing in the current technology. Summary of the Invention

[0005] In view of this, the present invention proposes a hybrid storage access scheduling method, system and medium for storage virtualization, which aims to solve the problems of frequent migration, access latency fluctuation and insufficient migration stability of logical block segments in the snapshot chain in the current storage virtualization environment due to the combined effects of short-term hot spot fluctuations, write-back pressure and additional write overhead during write access.

[0006] This invention proposes a hybrid storage access scheduling method for storage virtualization, comprising: Collect write access requests for target logical block segments in the target virtual volume, as well as the snapshot chain depth, hot flip number, and time interval of the target logical block segments; The urgency of the write access request is determined based on the write access request, and the snapshot chain depth, the number of hot flips, and the time interval are combined to determine whether the write access request meets the conditions for temporary acceleration processing. When it is determined that the write access request meets the temporary acceleration processing conditions, while keeping the main mapping relationship of the target logical block unchanged, a temporary acceleration mapping relationship of the target logical block is established in the high-speed storage layer, and the access request that hits the target logical block is directed to the temporary acceleration mapping relationship. The access frequency of the target logical block segment after the establishment of the temporary acceleration mapping relationship, the write-back status of the high-speed storage layer, and the amount of additional writes in the snapshot chain corresponding to the target logical block segment are collected. Based on the access frequency, the write-back status, and the amount of additional writes in the snapshot chain, it is determined whether the temporary acceleration mapping relationship meets the formal migration conditions. When it is determined that the temporary accelerated mapping relationship meets the formal migration conditions, the temporary accelerated mapping relationship is converted into a formal mapping relationship; otherwise, the temporary accelerated mapping relationship is revoked and the main mapping relationship remains unchanged.

[0007] Furthermore, the time interval is the time interval between the current write access request time and the last storage layer migration time of the target logical block segment; The temporary acceleration processing condition is: the urgency of the request reaches a preset level and meets any of the following: the snapshot chain depth is greater than a preset depth threshold, the number of heat flips is greater than a preset flip threshold, or the time interval is less than a preset cooling threshold.

[0008] Furthermore, when determining the urgency of a request based on the write access request, the process includes: obtaining the write type of the write access request and the current access latency of the target virtual volume; when the write type is metadata write or log write, and the current access latency is greater than a preset latency threshold, determining that the urgency of the request reaches the preset level; otherwise, determining that the urgency of the request is lower than the preset level.

[0009] Furthermore, establishing the temporary acceleration mapping relationship includes: In the high-speed storage layer, a temporary storage location corresponding to the target logical block segment is allocated, and a temporary acceleration mapping relationship is established between the target logical block segment and the temporary storage location. During a preset observation period, write access requests that hit the target logical block segment are directed to the temporary storage location, and the primary mapping relationship remains pointing to the storage layer before migration.

[0010] Furthermore, when it is determined that the write access request does not meet the temporary acceleration processing conditions, the main mapping relationship of the target logical block segment remains unchanged, and the write access request is directed to the storage layer corresponding to the main mapping relationship to perform the write; after the write is completed, the hot flip number and time interval of the target logical block segment are updated, and the updated hot flip number and time interval are used as input data for subsequent determination of whether the write access request meets the temporary acceleration processing conditions.

[0011] Furthermore, after directing the write access request to the storage layer corresponding to the main mapping relationship for writing, the number of write access requests that hit the target logical block segment within a preset statistical period is counted; when the number of write access requests reaches a preset threshold, the snapshot chain depth, hot flip number and time interval of the target logical block segment are collected again, and it is determined again whether the write access request meets the temporary acceleration processing conditions.

[0012] Furthermore, the formal migration conditions are as follows: the access frequency of the target logical block segment is continuously higher than the preset hotness threshold in multiple consecutive statistical periods, the write-back status of the high-speed storage layer is lower than the preset debt threshold, and the additional write volume of the snapshot chain is less than the preset write volume threshold.

[0013] Furthermore, the number of popularity reversals is the number of times the access frequency of the target logical block segment is compared with a preset popularity threshold within multiple consecutive statistical periods, and the statistical comparison result changes from being higher than the preset popularity threshold to being lower than the preset popularity threshold or from being lower than the preset popularity threshold to being higher than the preset popularity threshold.

[0014] Compared with existing technologies, the beneficial effects of this invention are as follows: By collecting data from the target logical block segment, the urgency of the request and the risk of block migration are jointly determined. When the temporary acceleration processing conditions are met, the main mapping relationship remains unchanged, and a temporary acceleration mapping relationship is established in the high-speed storage layer. This prioritizes the response efficiency of high-urgency write access without immediately changing the original data location, thereby avoiding frequent migrations, mapping oscillations, and additional write burdens caused by direct migration triggered by short-term hotspot fluctuations in existing technologies. At the same time, after the temporary acceleration mapping relationship is established, this solution further combines the access frequency, the write-back status of the high-speed storage layer, and the amount of additional write amplification in the snapshot chain to determine whether to convert to a formal mapping relationship. This ensures that the formal migration is based on continuous access demand, controllable write-back pressure, and acceptable overhead of the snapshot chain. This allows for the optimization of access latency, suppression of write amplification, and improvement of migration stability. It is suitable for write access scenarios under snapshot chain constraints and can effectively reduce the additional metadata processing pressure and write-back conflict risk caused by direct migration in the snapshot chain environment. It also improves the accuracy, stability, and overall operating efficiency of storage virtualization access scheduling in hybrid storage environments.

[0015] On the other hand, this application also provides a hybrid storage access scheduling system for storage virtualization, used to apply the above-mentioned hybrid storage access scheduling method for storage virtualization, including: The acquisition unit is configured to acquire write access requests of target logical block segments in the target virtual volume, as well as the snapshot chain depth, hot flip number and time interval of the target logical block segments; The determination unit is configured to determine the urgency of the write access request based on the write access request, and, in conjunction with the snapshot chain depth, the number of hot flips, and the time interval, determine whether the write access request meets the temporary acceleration processing conditions. The processing unit is configured to, when it is determined that the write access request meets the temporary acceleration processing conditions, establish a temporary acceleration mapping relationship for the target logical block segment in the high-speed storage layer while keeping the main mapping relationship of the target logical block segment unchanged, and guide the access request that hits the target logical block segment to the temporary acceleration mapping relationship. The adjustment unit is configured to collect the access frequency of the target logical block segment after the establishment of the temporary acceleration mapping relationship, the write-back status of the high-speed storage layer, and the additional write volume of the snapshot chain corresponding to the target logical block segment, and determine whether the temporary acceleration mapping relationship meets the formal migration conditions based on the access frequency, the write-back status, and the additional write volume of the snapshot chain. The adjustment unit is further configured to convert the temporary accelerated mapping relationship into a formal mapping relationship when it is determined that the temporary accelerated mapping relationship meets the formal migration conditions; otherwise, it cancels the temporary accelerated mapping relationship and keeps the main mapping relationship unchanged.

[0016] On the other hand, this application also provides a computer-readable storage medium in which computer-executable instructions stored in the computer-readable storage medium are executed by a processor of an electronic device, enabling the electronic device to perform the above-described method.

[0017] It is understandable that the hybrid storage access scheduling method, system and storage medium of the above-mentioned storage virtualization have the same beneficial effects, and will not be elaborated here. Attached Figure Description

[0018] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 A flowchart of a hybrid storage access scheduling method for storage virtualization provided in an embodiment of the present invention; Figure 2 A functional block diagram of a hybrid storage access scheduling system for storage virtualization provided in an embodiment of the present invention. Detailed Implementation

[0019] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specified, embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0020] In some embodiments of this application, see Figure 1 As shown, this application proposes a hybrid storage access scheduling method for storage virtualization, including: S100: Collect write access requests for the target logical block segment in the target virtual volume, as well as the snapshot chain depth, hot flip number, and time interval of the target logical block segment.

[0021] S200: Determine the urgency of the write access request based on the request, and combine this with the snapshot chain depth, hot flip count, and time interval to determine whether the write access request meets the conditions for temporary acceleration processing.

[0022] S300: When it is determined that the write access request meets the temporary acceleration processing conditions, while keeping the main mapping relationship of the target logical block segment unchanged, a temporary acceleration mapping relationship of the target logical block segment is established in the high-speed storage layer, and the access request that hits the target logical block segment is directed to the temporary acceleration mapping relationship.

[0023] S400: Collect the access frequency of the target logical block segment after establishing the temporary acceleration mapping relationship, the write-back status of the high-speed storage layer, and the amount of additional writes in the snapshot chain corresponding to the target logical block segment. Based on the access frequency, write-back status, and amount of additional writes in the snapshot chain, determine whether the temporary acceleration mapping relationship meets the formal migration conditions.

[0024] S500: When it is determined that the temporary accelerated mapping relationship meets the formal migration conditions, the temporary accelerated mapping relationship is converted into a formal mapping relationship. Otherwise, the temporary accelerated mapping relationship is revoked and the main mapping relationship remains unchanged.

[0025] Specifically, in this embodiment, in step S100, write access requests for target logical blocks in the target virtual volume, as well as the snapshot chain depth, hotspot flip count, and time interval of the target logical blocks, are collected. The target virtual volume can be a logical storage volume provided by the storage virtualization layer to a virtual machine, container, or upper-layer business system. The target logical block can be a block in the target virtual volume corresponding to the logical address range hit by the current write access request. The snapshot chain depth characterizes the number of levels traced upwards along the snapshot inheritance relationship of the data object containing the target logical block. A larger snapshot chain depth usually means that the logical block may trigger more redirect writes, metadata updates, or replica association processing during writing. The hotspot flip count characterizes the number of times the target logical block switches between high and low hotspots around a preset hotspot threshold within multiple consecutive statistical periods. This parameter does not simply reflect the access volume but is used to identify whether the target logical block has a situation of repeated short-term hotspot changes. The time interval is preferably the time interval between the current write access request and the last storage layer migration of the target logical block, used to characterize whether the target logical block is still in the migration cooling phase since the most recent migration. The above parameters can be obtained by the storage virtualization control layer through access logs, block mapping metadata, snapshot metadata, and migration record tables. Among them, the number of hot flips can be updated on a rolling basis according to a fixed statistical period, and the time interval can be calculated based on the current system time and the most recent migration time in the migration record.

[0026] In step S200, the urgency of the write access request is determined based on the request itself, and combined with the snapshot chain depth, hot flip count, and time interval, it is determined whether the write access request meets the conditions for temporary acceleration. The urgency of the request is not an abstract priority, but rather a scheduling metric reflecting whether the current write access request needs priority reduction in response latency. In some implementations, the write type of the write access request and the current access latency of the target virtual volume can be obtained. When the write type is metadata write or log write, and the current access latency is greater than a preset latency threshold, the urgency of the request is determined to reach a preset level; otherwise, the urgency of the request is determined to be lower than a preset level. This preset latency threshold can be determined based on the deployment scenario, business service level targets, or latency statistics during historical stable operation. For example, the average write latency, percentile latency, or management policy configuration value of the target virtual volume under normal load can be selected as the threshold source. Subsequently, the urgency of the request, along with the snapshot chain depth, hot flip count, and time interval, are used together to determine the conditions for temporary acceleration. For example, when the urgency of a request reaches a preset level, and the snapshot chain depth is greater than a preset depth threshold, the number of hot-spot flips is greater than a preset flip threshold, or the time interval is less than a preset cooling threshold, the current write access request can be determined to meet the conditions for temporary accelerated processing. These thresholds are not arbitrarily set. The preset depth threshold can be determined based on the upper limit of allowed snapshot chain processing complexity, the preset flip threshold can be determined based on the statistical distribution of historical hot-spot block flips, and the preset cooling threshold can be determined based on the stable observation period or empirical configuration after the storage layer migration is completed. In this way, migration is not triggered directly based solely on the level of hotness, but rather by identifying special write scenarios where "the request is urgent but it is not suitable to immediately proceed with formal migration."

[0027] In step S300, when a write access request is determined to meet the temporary acceleration processing conditions, a temporary acceleration mapping relationship for the target logical block is established in the high-speed storage layer while maintaining the primary mapping relationship of the target logical block. Access requests hitting the target logical block are then directed to this temporary acceleration mapping relationship. Here, the primary mapping relationship refers to the currently effective data location relationship of the target logical block, typically corresponding to the official storage location in the original performance storage layer or capacity storage layer. The temporary acceleration mapping relationship refers to the acceleration access mapping relationship that is only valid within a preset observation period, pointing to the storage location temporarily allocated for the target logical block in the high-speed storage layer. Maintaining the primary mapping relationship unchanged aims to prevent the current write access request from immediately modifying the official data location once it hits a short-term hotspot, thereby reducing metadata changes, chain redirects, and migration oscillations caused by frequent migrations. In implementation, the storage virtualization control layer can allocate a temporary storage location in the high-speed storage layer, prioritizing the writing of the current write access request and subsequent access requests hitting the target logical block to this temporary storage location, and maintaining the temporary acceleration mapping relationship from the target logical block to the temporary storage location in the mapping control table. The preset observation period can be set according to a number of consecutive statistical periods. The value can be determined according to the sensitivity requirements for identifying short-term hot spots. For example, it can be set to 3 to 8 consecutive statistical periods, thus providing an observation window for judging whether the accelerated behavior needs to be solidified into a formal migration.

[0028] In step S400, the access frequency of the target logical block segment after establishing the temporary accelerated mapping relationship, the write-back status of the high-speed storage layer, and the additional write volume of the snapshot chain corresponding to the target logical block segment are collected. Based on the access frequency, write-back status, and additional write volume of the snapshot chain, it is determined whether the temporary accelerated mapping relationship meets the formal migration conditions. The access frequency characterizes whether the target logical block segment maintains a high access status within a preset observation period, which can be obtained by statistically analyzing the number of accesses or writes to the target logical block segment within multiple consecutive statistical periods. The write-back status characterizes the data pressure in the high-speed storage layer that has not yet been flushed, merged, or formally written to disk, and can be characterized by at least one of the following: dirty data ratio, write-back queue length, and cumulative write-back data volume. The additional write volume of the snapshot chain characterizes the additional write overhead that may be introduced when the target logical block segment performs formal migration or formal placement adjustment under the current snapshot chain structure. This additional write overhead can be estimated by the number of redirected writes, the number of additional metadata updates, the number of associated block copies, or a combination thereof. The preferred conditions for formal migration are: the access frequency of the target logical block segment is consistently higher than a preset hotness threshold over multiple consecutive statistical periods; the write-back status of the high-speed storage layer is lower than a preset debt threshold; and the additional write volume of the snapshot chain is less than a preset write volume threshold. In a preferred embodiment, to avoid mixing different metering methods, the write-back status is uniformly represented by the proportion of the write-back data volume to the available space of the high-speed storage layer; the additional write volume of the snapshot chain is uniformly converted into the additional data write volume generated by the formal migration. The additional data write volume can be obtained by accumulating the data volume corresponding to the redirection write, the data volume corresponding to the snapshot metadata update, and the data volume corresponding to the associated block replication, and then comparing it with the preset write volume threshold to determine whether the formal migration conditions are met. The above thresholds can also be obtained by pre-configuration based on historical operating data, stress test results, or management strategies. Among them, the preset hotness threshold is used to identify whether a short-term hot spot has evolved into a persistent hot spot, the preset debt threshold is used to limit the continued solidification migration when the write-back pressure of the high-speed storage layer is high, and the preset write volume threshold is used to avoid performing formal migration when the additional overhead of the snapshot chain is too high.

[0029] In step S500, when it is determined that the temporary accelerated mapping relationship meets the formal migration conditions, the temporary accelerated mapping relationship is converted into a formal mapping relationship; otherwise, the temporary accelerated mapping relationship is revoked while the main mapping relationship remains unchanged. When converting the temporary accelerated mapping relationship into a formal mapping relationship, the mapping control module can update the temporary storage location in the high-speed storage layer to the formal storage location of the target logical block segment, and synchronously update the corresponding block mapping table, snapshot association information, and migration record information. When revoking the temporary accelerated mapping relationship, the temporary association between the target logical block segment and the temporary storage location is deleted, and only the original main mapping relationship is retained to continue providing subsequent access services. It can be seen that this embodiment does not confuse temporary acceleration processing with formal migration, but rather divides the two into a two-stage processing flow of "first alleviating urgent write access, and then determining whether to solidify the data location," so that the formal migration is based on the premise of continuous access demand and state allowance.

[0030] Specifically, in a scenario where a virtual machine hosts database services, if the target logical block segment corresponding to the database log block is located in a deep snapshot chain, and this target logical block segment fluctuates frequently around a hotness threshold in recent statistical periods, while the time since the last migration is relatively short, it indicates that the target logical block segment has an urgent write requirement that needs rapid response, but is not suitable for directly triggering a formal migration. In this case, this embodiment first establishes a temporary accelerated mapping relationship for the target logical block segment in the high-speed storage layer to prioritize handling current and short-term write access requests. Subsequently, during a preset observation period, it continues to observe whether the target logical block segment maintains a continuous high access state, and whether the write pressure pending in the high-speed storage layer and the large amount of additional writes in the snapshot chain are within a controllable range. If the observation results show that the target logical block segment has changed from a short-term fluctuating hotspot to a continuous hotspot and meets the conditions for undertaking a formal migration, the temporary accelerated mapping relationship is converted into a formal mapping relationship. If it is only a short-term sudden hotspot, the temporary accelerated mapping relationship is canceled to avoid unnecessary data placement adjustments.

[0031] It is understandable that by incorporating snapshot chain depth, hotspot flip count, and time interval into the temporary acceleration processing decision, the decision of "whether to accelerate the response" and "whether to immediately migrate" are separated from the same step. This allows for the priority suppression of invalid formal migrations caused by short-term hotspot fluctuations in snapshot chain write scenarios. This embodiment, by imposing phased constraints on the conditions for establishing temporary acceleration mapping relationships and the conditions for formal migration conversion, can first release the latency pressure of urgent write requests when the overhead of additional writes in the snapshot chain is still high, but the urgency of write access has already manifested. Then, based on the observation results, it is decided whether to solidify the data placement. This results in the simultaneous reduction of the false trigger rate of formal migration and the reduction of invalid occupation of the high-speed storage layer. This effect is not naturally brought about by simply increasing the decision parameters, but rather comes from the phased processing mechanism of "temporary acceleration - delayed confirmation - then conversion to formal mapping".

[0032] In some embodiments of this application, the time interval is the time interval between the current write access request and the last storage layer migration time of the target logical block segment. The temporary acceleration processing condition is: the urgency of the request reaches a preset level, and any of the following is met: the snapshot chain depth is greater than a preset depth threshold, the number of hot flips is greater than a preset flip threshold, or the time interval is less than a preset cooling threshold.

[0033] In some embodiments of this application, determining the urgency of a write access request includes: obtaining the write type of the write access request and the current access latency of the target virtual volume. When the write type is metadata write or log write, and the current access latency is greater than a preset latency threshold, the urgency of the request is determined to have reached a preset level. Otherwise, the urgency of the request is determined to be lower than the preset level.

[0034] Specifically, in this embodiment, the time interval is used to reflect whether the target logical block segment is still in a stable observation phase since the most recent formal migration. The time corresponding to the current write access request can be the system time recorded when the storage virtualization control layer receives the write access request. The time of the last storage layer migration can be read from the block migration record table, mapping relationship change log, or migration task completion log. Preferably, the above time is generated by the clock module under the same control node or the same time synchronization domain to reduce statistical bias caused by time source differences. When the time interval is small, it indicates that the target logical block segment has just undergone storage layer adjustment. If a formal migration is performed again at this time, it is easy to trigger continuous mapping changes, data migration, or additional metadata updates. Therefore, it is used as one of the inputs for determining the temporary acceleration processing condition. The temporary acceleration processing condition is not determined based on a single heat index, but based on the joint determination result of the request urgency and the migration restricted state. The snapshot chain depth is used to characterize the hierarchical position of the target logical block segment in the snapshot inheritance relationship. The larger the snapshot chain depth, the more complex the snapshot association relationship that needs to be synchronized during the formal migration. The number of heat reversals is used to characterize the frequency of high-low switching of a target logic block relative to a preset heat threshold within multiple consecutive statistical periods. It can be obtained by comparing the access frequency in each statistical period with the preset heat threshold and counting the number of times the comparison result changes from above the threshold to below the threshold or from below the threshold to above the threshold.

[0035] When determining the urgency of a write access request, the write type can be determined by the request tag information in the file system, volume management module, log management module, or virtualization input / output stack. The current access latency can be obtained by the average write latency, percentile write latency, or response time of the most recent write requests within a preset latency statistics window for the target virtual volume. The preset latency threshold can be obtained based on the baseline latency during historical stable operation, service level targets, or system initialization configuration. The preset depth threshold, preset flip threshold, and preset cooling threshold can be determined based on snapshot processing overhead statistics, historical distribution of hotspot fluctuations, and post-migration stabilization test results. In a preferred embodiment, the current access latency is obtained using the percentile response latency of write access requests within the preset latency statistics window, preferably the 95th percentile write latency. The preset latency statistics window can be set from 200 milliseconds to 2 seconds. Using percentile response latency instead of single response time helps reduce the impact of occasional jitter on the request urgency determination result.

[0036] In one implementation, stress tests can be performed on typical workloads during the initial deployment phase to obtain latency variation curves under different snapshot chain depths, hotspot flipping intensities, and migration intervals. Based on this, thresholds that keep false trigger rates and missed trigger rates within acceptable ranges can be selected. It should be noted that reaching a preset urgency level does not automatically mean that a formal migration must be performed; rather, it indicates that the current request has a priority in reducing response latency. Only when this urgency and the migration constraint both occur simultaneously will the temporary acceleration processing branch be initiated.

[0037] Specifically, in scenarios where database virtual volumes handle transaction processing, when a target logical block corresponds to a database log area and is located in a deep snapshot chain, if a write access request is detected as a log write and the current access latency of the target virtual volume has exceeded a preset latency threshold, the urgency of the request can be determined to have reached a preset level. If the target logical block frequently crosses a preset heat threshold in several consecutive statistical periods, or if the time since the last storage layer migration has not exceeded a preset cooling threshold, the write access request can be determined to meet the conditions for temporary acceleration processing. Therefore, instead of directly modifying the master mapping relationship or immediately initiating a formal migration, a temporary acceleration process is initiated first, thereby reducing log write waiting time while avoiding triggering a new formal migration due to recent migration completion or short-term hotspot fluctuations.

[0038] Understandably, by simultaneously incorporating request urgency and migration constraint status into the triggering logic of temporary acceleration conditions, it is possible to identify special scenarios that traditional hot migration strategies struggle to handle accurately. These scenarios involve target logic blocks that urgently require write operations to reduce latency but lack the stable conditions for immediate formal migration. This embodiment further clarifies when the first phase should be triggered and limits the triggering premise to the coexistence of business urgency and migration risk. This reduces both the excessive acceleration of ordinary write requests and the probability of misjudging short-term, non-critical fluctuations as priority targets, allowing the acceleration resources of the high-speed storage layer to be more concentrated on write requests that truly impact business latency.

[0039] In some embodiments of this application, establishing a temporary acceleration mapping relationship includes: allocating a temporary storage location corresponding to the target logical block segment in the high-speed storage layer, and establishing a temporary acceleration mapping relationship between the target logical block segment and the temporary storage location. During a preset observation period, write access requests that hit the target logical block segment are directed to the temporary storage location, while the primary mapping relationship remains pointing to the storage layer before migration.

[0040] Specifically, in this embodiment, the temporary accelerated mapping relationship refers to a time-limited, effective accelerated access association established for the target logical block segment in the high-speed storage layer without changing the existing formal data location relationship. The high-speed storage layer can be composed of solid-state drives, non-volatile high-speed media, or other storage resources with lower access latency compared to the pre-migration storage layer. The temporary storage location can be storage space allocated in the high-speed storage layer at the block level, page level, or preset mapping unit level. The primary mapping relationship refers to the mapping relationship between the target logical block segment and its formal storage location in the pre-migration storage layer. This primary mapping relationship remains valid in this embodiment and is not immediately changed due to the establishment of the temporary accelerated mapping relationship.

[0041] In practical implementation, the space management module can first detect the available space in the high-speed storage layer, and then allocate temporary storage locations for the target logical block segment according to the block length, alignment rules, and the minimum allocation unit of the high-speed storage layer. The block length can be determined based on the logical address range hit by the current write access request, or it can be determined based on a pre-set fixed block size. The alignment rules are preferably consistent with the physical page boundaries, erase / write unit boundaries, or cache management granularity of the high-speed storage layer to reduce cross-page writes and additional write amplification.

[0042] After allocation, the mapping management module writes information such as the target logical block identifier, temporary storage location identifier, establishment time, and failure conditions into the temporary mapping table to form a temporary accelerated mapping relationship. Failure conditions can include the expiration of the preset observation period, the fulfillment of formal migration conditions, or failure to meet formal migration conditions after the preset observation period expires. The preset observation period limits the effective duration of the temporary accelerated mapping relationship and can be set by time length or the number of consecutive statistical periods, such as 3 to 8 consecutive statistical periods or a continuous observation window of 100 milliseconds to 2 seconds. Its value can be determined based on the sensitivity of short-term hotspot identification, the frequency of business load changes, and the capacity pressure of the high-speed storage layer. The preset observation period is not simply a delay in processing time, but rather provides a necessary behavioral observation window for subsequent judgment on whether the target logical block should transition from the temporary accelerated state to the formal mapping state. During this observation period, the request parsing module preferably queries the temporary mapping table first. If a valid temporary accelerated mapping relationship is detected for the target logical block, the write access request that hits the target logical block is written to the temporary storage location, while the main mapping relationship remains in its original state in the main mapping table and is not updated to the location corresponding to the high-speed storage layer. After the write operation is complete, runtime data such as the number of writes, cumulative write volume, number of temporary mapping hits, and duration after creation can be recorded for later use in determining whether the formal migration conditions are met. If necessary, valid and version flags can be set for the temporary storage location to prevent old temporary data from being incorrectly referenced after the temporary accelerated mapping relationship is revoked. Therefore, the temporary accelerated mapping relationship is essentially an additional mapping relationship, rather than an immediate replacement of the main mapping relationship.

[0043] Specifically, once a target logical block segment meets the conditions for temporary acceleration processing, a free area that meets the capacity and alignment requirements can be selected from the high-speed storage layer as a temporary storage location. The correspondence between the target logical block segment and this temporary storage location is then registered in the temporary mapping table, and the preset observation period end time or the corresponding observation period count is written. Subsequently, during the preset observation period, when a new write access request hits the target logical block segment, the scheduling control module, based on the query result of the temporary mapping table, directly writes the write access request to the temporary storage location without modifying the original formal location record of the target logical block segment in the main mapping table. Thus, on the one hand, the high-speed storage layer can quickly handle the high-pressure writes of the current stage; on the other hand, the original formal mapping basis of the target logical block segment in the storage layer before migration is preserved, allowing for flexible decisions based on subsequent observation results regarding whether to switch to formal mapping or directly cancel the temporary acceleration mapping relationship.

[0044] Understandably, by combining the processing method of "temporary storage location allocation + additional mapping establishment + main mapping remaining unchanged," the high-speed storage layer can absorb high-pressure write accesses in an additional manner without immediately triggering the chained metadata adjustments and data placement confirmations associated with the formal mapping switch. This embodiment makes the temporary acceleration phase a parallel mapping mechanism decoupled from the formal migration, so that even if the formal migration conditions are not met later, there is no need to perform a formal mapping rollback. The original main mapping state can be restored simply by canceling the temporary acceleration mapping relationship. This reduces the formal mapping jitter and mapping rollback costs caused by short-term hotspot false triggers.

[0045] In some embodiments of this application, when a write access request is determined not to meet the temporary acceleration processing conditions, the primary mapping relationship of the target logical block segment remains unchanged, and the write access request is directed to the storage layer corresponding to the primary mapping relationship for write execution. After the write is completed, the number of hot flips and the time interval of the target logical block segment are updated, and the updated number of hot flips and the time interval are used as input data for subsequent determination of whether the write access request meets the temporary acceleration processing conditions.

[0046] In some embodiments of this application, after the write access request is directed to the storage layer corresponding to the main mapping relationship for writing, the number of write access requests hitting the target logical block segment within a preset statistical period is counted. When the number of write access requests reaches a preset threshold, the snapshot chain depth, hot flip number, and time interval of the target logical block segment are collected again, and it is determined again whether the write access request meets the temporary acceleration processing conditions.

[0047] Specifically, in this embodiment, when a write access request is determined not to meet the temporary acceleration processing conditions, it means that the current request has not yet simultaneously met the combined triggering conditions of "the request has sufficient urgency" and "the target logical block segment has constraints that make it unsuitable for immediate formal migration." Therefore, it does not enter the temporary acceleration processing branch of the high-speed storage layer, but instead executes the write along the existing path corresponding to the main mapping relationship. The storage layer corresponding to the main mapping relationship can be a performance storage layer or a capacity storage layer, the specific type of which depends on the formal placement status of the target logical block segment at the current moment. After completing the temporary acceleration processing condition determination, the request scheduling module can directly query the main mapping relationship corresponding to the target logical block segment, obtain the target storage layer and formal storage location pointed to by the main mapping relationship, and direct the write access request to the target storage layer for write execution. Here, "direction" includes mapping the logical address of the write access request to the physical address corresponding to the main mapping relationship, completing the write submission according to the current storage layer's write protocol, and returning the write result after the write is completed. Unlike the temporary acceleration branch, this embodiment does not allocate temporary storage locations, establish additional mapping relationships, or trigger temporary observation periods. Instead, it maintains the existing data placement stability of the target logical block segment based on the main mapping relationship. After the write operation is completed, the hot flip count and time interval of the target logical block segment are updated. The hot flip count can be updated based on the switching of the target logical block segment's access frequency relative to a preset hot threshold over multiple consecutive statistical periods. Preferably, the update of the hot flip count is completed at the end of the statistical period. After a single write operation, the cumulative access frequency of the target logical block segment within the current statistical period can be updated first. After the current statistical period ends, the hot flip count is determined based on whether a switch occurs in the comparison results of adjacent statistical periods relative to the preset hot threshold. If the write operation causes the access frequency within the current statistical period to cross the preset hot threshold, a new flip count may be generated during comparisons in subsequent periods. The time interval is preferably the time interval between the current write access request time and the time of the target logical block segment's last storage layer migration. If the current write operation does not trigger a formal migration, the time of the last storage layer migration usually remains unchanged; therefore, the time interval can be obtained by recalculating the difference between the current system time and the last storage layer migration time.

[0048] To avoid immediately re-executing the complete temporary acceleration processing decision after each normal write, a further decision-making mechanism based on the number of write access requests is set up. Specifically, after directing the write access request to the storage layer corresponding to the main mapping relationship for write execution, the number of write access requests hitting the target logical block segment within a preset statistical period is counted. When the number of write access requests reaches a preset threshold, the snapshot chain depth, hotspot flip count, and time interval of the target logical block segment are re-collected, and the write access request is re-determined to meet the temporary acceleration processing conditions. The preset statistical period can be set according to a fixed time window, such as 100 milliseconds to 1 second, or according to a fixed scheduling period, and its value can be determined according to the response requirements of the hotspot formation speed. The preset threshold can be determined based on historical access trajectory statistics, block hit distribution during peak business periods, or offline stress test results. In this way, this embodiment establishes a closed-loop processing flow of "normal write - status update - re-evaluation after reaching the aggregation threshold". Re-collecting the snapshot chain depth, number of hot flips, and time interval, and then re-determining whether the temporary acceleration processing conditions are met, does not mean that the previous determination is negated. Rather, it means that after the target logical block segment access status has undergone cumulative changes, it is allowed to re-enter the determination process of the temporary acceleration processing conditions.

[0049] Specifically, if a target logical block initially does not meet the conditions for temporary acceleration processing, write access requests can be written to the corresponding official location in the storage layer before migration based on the primary mapping relationship. After writing, the hot flip count and time interval of the target logical block are updated. If, in a subsequent preset statistical period, the target logical block is continuously hit by new write access requests, and the cumulative number of write access requests reaches a preset threshold, the snapshot chain depth, hot flip count, and time interval of the target logical block are re-collected, and the temporary acceleration processing condition determination is performed again. If the hot flip count increases due to access aggregation, or the time interval meets the new determination state due to the passage of cooling time, the target logical block can regain the opportunity to enter the temporary acceleration processing branch. Thus, even if temporary acceleration processing is not triggered initially, the scheduling decision can still be corrected in a timely manner based on subsequent access evolution.

[0050] Understandably, by employing a continuous control method of "writing the main mapping relationship + continuously updating the state variable + re-judging after reaching the aggregation threshold," the evolution of the target logical block segment from a normal write state to a high-value accelerated object state can be tracked without additionally consuming high-speed storage layer resources. This embodiment supplements the dynamic backflow mechanism when temporary acceleration processing is not triggered, ensuring that unaccelerated write access requests do not remain permanently on the low-priority processing path, but are reintegrated into the core judgment process when access aggregation reaches a certain level, thereby improving the overall scheduling mechanism's ability to capture medium-intensity, persistent hotspots.

[0051] In some embodiments of this application, the formal migration conditions are: the access frequency of the target logical block segment is continuously higher than a preset hotness threshold in multiple consecutive statistical periods, the write-back status of the high-speed storage layer is lower than a preset debt threshold, and the amount of additional writes in the snapshot chain is less than a preset write volume threshold.

[0052] In some embodiments of this application, the number of heat reversals is the number of times the access frequency of the target logic block segment is compared with a preset heat threshold within multiple consecutive statistical periods, and the statistical comparison result changes from higher than the preset heat threshold to lower than the preset heat threshold or from lower than the preset heat threshold to higher than the preset heat threshold.

[0053] Specifically, in this embodiment, the formal migration condition is used to characterize whether the target logical block has changed from a state of "suitable for temporary acceleration" to a state of "suitable for formal data location change". The number of hotspot reversals is used to characterize whether the access hotspot of the target logical block fluctuates repeatedly over multiple consecutive statistical periods. These two conditions act at different stages: the former determines whether the temporary acceleration mapping relationship can be solidified into a formal mapping relationship, while the latter reflects whether the target logical block has the characteristic of short-term hotspot recurrence before entering the temporary acceleration processing judgment. In the formal migration condition, the access frequency is preferably the number of write access requests hitting the target logical block within a single statistical period. It can also be characterized by a combination of request count and written data volume according to actual needs, but a consistent standard should be adopted in the same implementation. Multiple consecutive statistical periods are preferably 3 to 10 consecutive statistical periods. The preset hotspot threshold can be determined based on the access frequency distribution of hotspot blocks in historical operating data, stress test results, or statistical results during peak business periods. "Continuously higher than" means that the access frequency corresponding to each statistical period is higher than the preset hotspot threshold over multiple consecutive statistical periods, rather than only being higher than the threshold in some periods. The write-back status of the high-speed storage layer is used to characterize the data pressure that the high-speed storage layer has not yet completed write-back, merging, or flushing. It can be characterized by the amount of data to be written back, the proportion of available space occupied by the data to be written back, the length of the write-back queue, or a combination of these parameters.

[0054] In a preferred embodiment, the proportion of data to be written back to the available space of the high-speed storage layer can be uniformly used as the write-back status, and this proportion can be compared with a preset debt threshold. When the proportion is lower than the preset debt threshold, it indicates that the high-speed storage layer still has the capacity to accommodate data after formal migration. Snapshot chain additional write amplification is used to characterize the additional write overhead that may be introduced when formally migrating the target logical block segment to the high-speed storage layer under the current snapshot chain structure. This additional write overhead can be estimated by the number of redirected writes, the number of snapshot metadata updates, the number of associated block copies, or the data write volume corresponding to these overheads. The preset write amplification threshold can be determined based on the system's acceptable additional write limit, media lifetime protection strategy, or offline migration test results. In this embodiment, the number of hotspot reversals is the number of times the access frequency of the target logical block segment changes from higher than the preset hotspot threshold to lower than the preset hotspot threshold or vice versa within multiple consecutive statistical periods.

[0055] Specifically, the access frequency of the target logic block segment can be collected within each statistical period according to a fixed length. Then, the access frequency is compared with the same preset heat threshold period by period to form a sequence of high-heat or low-heat states. Between two adjacent statistical periods, if the state changes from high-heat to low-heat, the heat flip count increases by 1. If the state changes from low-heat to high-heat, the heat flip count also increases by 1. If the state remains unchanged, the heat flip count does not increase. To avoid unclear comparison boundaries, in one embodiment, the case where the access frequency is equal to the preset heat threshold can be regarded as a continuation of the state of the previous statistical period. If it is the first statistical period and the access frequency is equal to the preset heat threshold, it can be regarded as a low-heat state by default. Through the above processing, the false count caused by threshold boundary jitter can be reduced. The heat flip count essentially reflects the stability of the access heat of the target logic block segment: when the heat flip count is high, it indicates that the access frequency of the target logic block segment fluctuates frequently around the preset heat threshold, and it is usually not advisable to directly perform formal migration. When the number of popularity flips is low and the access frequency is consistently higher than the preset popularity threshold, it indicates that the target logical block segment is closer to a continuous hot spot state.

[0056] Specifically, in a certain implementation scenario, the statistical period is set to 200 milliseconds. The access frequency of the target logical block segment within five consecutive statistical periods is 12, 14, 8, 13, and 15 times respectively. The preset heat threshold is 10 times. Therefore, the first two statistical periods are in a high-heat state, the third statistical period becomes a low-heat state, and the fourth and fifth statistical periods return to a high-heat state, resulting in two heat reversals. If another target logical block segment has an access frequency higher than 10 times within five consecutive statistical periods, and the write-back status of the high-speed storage layer is lower than the preset debt threshold, and the estimated write-amount of the snapshot chain based on the current snapshot chain depth and associated metadata update volume is lower than the preset write-amount threshold, then the target logical block segment can be determined to meet the formal migration conditions, thereby converting the temporary acceleration mapping relationship into a formal mapping relationship. This implementation method can distinguish between target logical blocks that are "continuously high-heat and the system has the capacity to handle them" and target logical blocks that are "short-term fluctuating but unstable."

[0057] Understandably, by independently characterizing the block popularity fluctuations through the number of popularity flips, it is possible to simultaneously identify "whether migration is worthwhile" and "whether migration is suitable" within the same scheduling system. This embodiment further improves the accuracy of the second-stage conversion judgment, which can not only suppress the misconversion of short-term hot spots to formal mapping, but also avoid premature data fixation when the write-back pressure of the high-speed storage layer is too high or the additional write overhead of the snapshot chain is too large. Thus, while maintaining the success rate of formal migration, it reduces the probability of rollback and the additional write burden after formal migration.

[0058] Based on another preferred embodiment described above, see [link to preferred embodiment]. Figure 2 As shown, this embodiment provides a hybrid storage access scheduling system for storage virtualization, used to apply the above-described hybrid storage access scheduling method for storage virtualization, including: The acquisition unit is configured to acquire write access requests for target logical block segments in the target virtual volume, as well as the snapshot chain depth, hot flip count, and time interval of the target logical block segments.

[0059] The determination unit is configured to determine the urgency of the write access request based on the write access request, and, in conjunction with the snapshot chain depth, the number of hot flips, and the time interval, determine whether the write access request meets the conditions for temporary acceleration processing.

[0060] The processing unit is configured to, when it is determined that a write access request meets the conditions for temporary acceleration processing, establish a temporary acceleration mapping relationship for the target logical block segment in the high-speed storage layer while keeping the main mapping relationship of the target logical block segment unchanged, and guide the access request that hits the target logical block segment to the temporary acceleration mapping relationship.

[0061] The adjustment unit is configured to collect the access frequency of the target logical block segment after the establishment of the temporary acceleration mapping relationship, the write-back status of the high-speed storage layer, and the amount of additional writes in the snapshot chain corresponding to the target logical block segment. Based on the access frequency, write-back status, and the amount of additional writes in the snapshot chain, it determines whether the temporary acceleration mapping relationship meets the formal migration conditions.

[0062] The adjustment unit is also configured to convert a temporary accelerated mapping relationship into a formal mapping relationship when it is determined that the temporary accelerated mapping relationship meets the formal migration conditions. Otherwise, it cancels the temporary accelerated mapping relationship and keeps the primary mapping relationship unchanged.

[0063] On the other hand, this application also provides a computer-readable storage medium in which the electronic device can perform the above-described method when the computer-executable instructions stored in the computer-readable storage medium are executed by the processor of the electronic device.

[0064] It is understandable that the hybrid storage access scheduling method, system and storage medium of the above-mentioned storage virtualization have the same beneficial effects, and will not be elaborated here.

[0065] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A hybrid storage access scheduling method for storage virtualization, characterized in that, include: Collect write access requests for target logical block segments in the target virtual volume, as well as the snapshot chain depth, hot flip number, and time interval of the target logical block segments; The urgency of the write access request is determined based on the write access request, and the snapshot chain depth, the number of hot flips, and the time interval are combined to determine whether the write access request meets the conditions for temporary acceleration processing. When it is determined that the write access request meets the temporary acceleration processing conditions, while keeping the main mapping relationship of the target logical block unchanged, a temporary acceleration mapping relationship of the target logical block is established in the high-speed storage layer, and the access request that hits the target logical block is directed to the temporary acceleration mapping relationship. The access frequency of the target logical block segment after the establishment of the temporary acceleration mapping relationship, the write-back status of the high-speed storage layer, and the amount of additional writes in the snapshot chain corresponding to the target logical block segment are collected. Based on the access frequency, the write-back status, and the amount of additional writes in the snapshot chain, it is determined whether the temporary acceleration mapping relationship meets the formal migration conditions. When it is determined that the temporary accelerated mapping relationship meets the formal migration conditions, the temporary accelerated mapping relationship is converted into a formal mapping relationship; Otherwise, cancel the temporary acceleration mapping relationship and keep the main mapping relationship unchanged; The time interval is the time interval between the current write access request time and the last storage layer migration time of the target logical block segment; The temporary acceleration processing condition is: the urgency of the request reaches a preset level and meets any of the following: the snapshot chain depth is greater than a preset depth threshold, the number of heat flips is greater than a preset flip threshold, or the time interval is less than a preset cooling threshold. The formal migration conditions are as follows: the access frequency of the target logical block segment is continuously higher than the preset hotness threshold in multiple consecutive statistical periods, the write-back status of the high-speed storage layer is lower than the preset debt threshold, and the additional write volume of the snapshot chain is less than the preset write volume threshold.

2. The hybrid storage access scheduling method for storage virtualization according to claim 1, characterized in that, When determining the urgency of a write access request, the process includes: obtaining the write type of the write access request and the current access latency of the target virtual volume; when the write type is metadata write or log write, and the current access latency is greater than a preset latency threshold, determining that the urgency of the request reaches the preset level; otherwise, determining that the urgency of the request is lower than the preset level.

3. The hybrid storage access scheduling method for storage virtualization according to claim 1, characterized in that, Establishing the temporary acceleration mapping relationship includes: In the high-speed storage layer, a temporary storage location corresponding to the target logical block segment is allocated, and a temporary acceleration mapping relationship is established between the target logical block segment and the temporary storage location. During a preset observation period, write access requests that hit the target logical block segment are directed to the temporary storage location, and the primary mapping relationship remains pointing to the storage layer before migration.

4. The hybrid storage access scheduling method for storage virtualization according to claim 1, characterized in that, When it is determined that the write access request does not meet the temporary acceleration processing conditions, the main mapping relationship of the target logical block segment remains unchanged, and the write access request is directed to the storage layer corresponding to the main mapping relationship to perform the write; after the write is completed, the hot flip count and time interval of the target logical block segment are updated, and the updated hot flip count and time interval are used as input data for subsequent determination of whether the write access request meets the temporary acceleration processing conditions.

5. The hybrid storage access scheduling method for storage virtualization according to claim 4, characterized in that, After directing the write access request to the storage layer corresponding to the main mapping relationship for write execution, the number of write access requests that hit the target logical block segment within a preset statistical period is counted. When the number of write access requests reaches a preset threshold, the snapshot chain depth, hot flip count, and time interval of the target logical block segment are collected again, and it is determined again whether the write access request meets the temporary acceleration processing conditions.

6. The hybrid storage access scheduling method for storage virtualization according to claim 1, characterized in that, The number of popularity reversals is the number of times the access frequency of the target logical block segment is compared with a preset popularity threshold within multiple consecutive statistical periods, and the statistical comparison result changes from being higher than the preset popularity threshold to being lower than the preset popularity threshold or from being lower than the preset popularity threshold to being higher than the preset popularity threshold.

7. A hybrid storage access scheduling system for storage virtualization, used to apply the hybrid storage access scheduling method for storage virtualization as described in any one of claims 1-6, characterized in that, include: The acquisition unit is configured to acquire write access requests of target logical block segments in the target virtual volume, as well as the snapshot chain depth, hot flip number and time interval of the target logical block segments; The determination unit is configured to determine the urgency of the write access request based on the write access request, and, in conjunction with the snapshot chain depth, the number of hot flips, and the time interval, determine whether the write access request meets the temporary acceleration processing conditions. The processing unit is configured to, when it is determined that the write access request meets the temporary acceleration processing conditions, establish a temporary acceleration mapping relationship for the target logical block segment in the high-speed storage layer while keeping the main mapping relationship of the target logical block segment unchanged, and guide the access request that hits the target logical block segment to the temporary acceleration mapping relationship. The adjustment unit is configured to collect the access frequency of the target logical block segment after the establishment of the temporary acceleration mapping relationship, the write-back status of the high-speed storage layer, and the additional write volume of the snapshot chain corresponding to the target logical block segment, and determine whether the temporary acceleration mapping relationship meets the formal migration conditions based on the access frequency, the write-back status, and the additional write volume of the snapshot chain. The adjustment unit is further configured to convert the temporary accelerated mapping relationship into a formal mapping relationship when it is determined that the temporary accelerated mapping relationship meets the formal migration conditions; otherwise, it cancels the temporary accelerated mapping relationship and keeps the main mapping relationship unchanged.

8. A computer-readable storage medium, characterized in that, When the computer-executable instructions stored in the computer-readable storage medium are executed by the processor of the electronic device, the electronic device is capable of performing the method as described in any one of claims 1 to 6.

Images