WO2016165542A1 - Method for analyzing cache hit rate, and device - Google Patents

Abstract

The present application provides a method for analyzing a cache hit rate, and a device. In the present application, data of a low cache hit rate is obtained according to the total number of times of data cache read failure events, and a cause of the low cache hit rate can be effectively determined according the total quantity of the data read failure events for the low cache hit rate and the other cache events, thereby improving the quality of cache codes according to the cause, and improving the performance of the whole system.

Description

Method and device for analyzing cache hit rate Technical field

The present application relates to the field of communications and computers, and in particular, to a method and device for analyzing cache hit ratio.

Background technique

Cache is a system for temporarily storing data. The difference between cache and database (DB), file system and other storage systems is that the capacity is limited, the price is expensive, and the power loss is lost, but the access speed is fast and the writing speed is fast. Cache is often used as a data buffer for the system, putting frequently accessed data into the cache, avoiding multiple visits to the database, or slow access to the file system. One of the core metrics in the cache system is the hit rate, which is equal to the number of times the read request returns to exist/the total number of reads. Among them, a hit refers to a hit when reading a certain data, and a hit if the data exists; a miss means that when a certain data is read, if the data does not exist, it is a non-hit.

The cache hit rate is closely related to the user's usage logic. When the hit rate is low, the performance of the system cannot be improved, so the hit rate directly affects the response time and overall throughput of the system. As for how to improve the hit rate, there is currently no effective quantitative method. The existing judgment method is basically based on the developer's understanding of the code, the familiarity of the business structure, and the reason for the low judgment of the man-made rate. It can only be judged by human experience and there is no tool to rely on.

Summary of the invention

The purpose of the present application is to provide a method and device for analyzing cache hit ratio, which can obtain an accurate reason for low cache hit ratio.

In view of this, the present application provides a method for analyzing a cache hit ratio, which includes:

Obtaining an event record of the same type of data whose cache read failure event is higher than the first preset threshold;

The reason why the data cache hit rate is low is determined according to the cumulative number of read failure events and other cache events of each data in the event record of the acquired similar data.

Further, the other cache events include one or any combination of a delete event, a write event, and a retirement event.

Further, the reason for the low data hit hit rate is determined according to the cumulative number of read failure events and other cache events of each data in the event record of the acquired similar data, including:

When the cumulative number of deletion events and read failure events of a certain data is greater than a second preset threshold, and the difference between the cumulative number of deletion events and the cumulative number of read failure events is less than a third preset threshold, determining The reason for the low data cache hit rate is that the data is actively deleted.

Further, the reason for the low data hit hit rate is determined according to the cumulative number of read failure events and other cache events of each data in the event record of the acquired similar data, including:

When the accumulated number of read failure events of a certain data is greater than a fourth preset threshold, and the difference between the cumulative number of read failure events and the accumulated number of write events is greater than a fifth preset threshold, determining that the data cache hits The reason for the low rate is that the accessed data does not exist.

Further, the reason for the low data hit hit rate is determined according to the cumulative number of read failure events and other cache events of each data in the event record of the acquired similar data, including:

When the cumulative number of the defunct event and the read failure event of a certain data is greater than the sixth preset threshold, and the difference between the cumulative number of the culling event and the accumulated number of the read failure event is less than the seventh preset threshold, determining the The reason for the low data cache hit rate is that the cache capacity is insufficient.

Further, the reason for the low data hit hit rate is determined according to the cumulative number of read failure events and other cache events of each data in the event record of the acquired similar data, including:

When the cumulative number of the write event and the read failure event of a certain data is less than the eighth preset threshold, and the difference between the accumulated number of the write event and the accumulated number of the read failure event is less than the ninth preset threshold, then determining The reason why the data cache hit rate is low is that the accessed data is not hot.

Further, the event record of the same type of data that the sum of the cache read failure events is higher than the first preset threshold is obtained, including:

Get the cumulative number of cache read failure events and other cache events for each data;

Filtering data from all the data that the cumulative number of read failure events is greater than the tenth preset threshold;

Each of the filtered data is assigned to a corresponding class according to a preset classification criterion;

Calculate the total number of cache read failure events for each type of data;

Obtaining an event record of the same type of data whose sum of cache read failure events is higher than a first preset threshold, wherein the event record of the same type of data includes a read failure event of each data in the type of data and other cache events The cumulative quantity.

Further, the cumulative number of cache read failure events and other cache events for each data is obtained, including:

Obtaining a cache read failure event or other cache event for each data through a callback interface, the callback interface being disposed on an event path of the cache code;

Calculates the cumulative number of cache read failure events and other cache events for each data.

Further, the preset classification criteria include classification according to a function of data or a unique identifier of data.

Further, the reason for the low data hit hit rate is determined according to the cumulative number of read failure events and other cache events of each data in the event record of the acquired similar data, including:

All the data in the event record of the acquired similar data is generated according to the size of the cumulative number of read failure events and other cache events;

The reason for the low cache hit ratio is determined based on the chart.

According to another aspect of the present application, there is also provided an apparatus for cache hit ratio analysis, comprising:

a first device, configured to acquire an event record of the same type of data whose cache read failure event is higher than a first preset threshold;

a second device for reading each data in the event record according to the acquired similar data The cumulative number of failed events and other cached events determines why the data cache hit rate is low.

Further, the other cache events include one or any combination of a delete event, a write event, and a retirement event.

Further, the second device is configured to: when the cumulative number of deletion events and read failure events of a certain data is greater than a second preset threshold, and the difference between the cumulative number of deletion events and the cumulative number of read failure events If the value is less than the third preset threshold, it is determined that the data cache hit rate is low because the data is actively deleted.

Further, the second device is configured to: when the cumulative number of read failure events of a certain data is greater than a fourth preset threshold, and the difference between the cumulative number of read failure events and the accumulated number of write events is greater than The five preset thresholds determine that the data cache hit rate is low because the accessed data does not exist.

Further, the second device is configured to: when the cumulative number of the elimination events and the read failure events of the certain data are greater than the sixth preset threshold, and the difference between the cumulative number of the eliminated events and the accumulated number of the read failure events If the value is less than the seventh preset threshold, it is determined that the data cache hit rate is low because the cache capacity is insufficient.

Further, the second device is configured to: when the cumulative number of write events and read failure events of a certain data is less than an eighth preset threshold, and the cumulative number of write events and the cumulative number of read failure events If the difference is less than the ninth preset threshold, it is determined that the data cache hit rate is low because the accessed data has no hotspot.

Further, the first device includes:

The first module is configured to obtain a cumulative number of cache read failure events and other cache events for each data;

The first two modules are configured to filter, from all data, data that the cumulative number of read failure events is greater than a tenth preset threshold;

The first three modules are configured to classify each of the filtered data into corresponding classes according to a preset classification criterion;

The first four modules are used to calculate the total number of cache read failure events of each type of data;

a first fifth module, configured to obtain an event record of the same type of data whose cache read failure event is higher than a first preset threshold, wherein the event record of the same type data includes reading of each data in the type of data The cumulative number of failed events and other cached events.

Further, the first module includes:

a first one unit for obtaining a cache read failure event or other cache event of each data through a callback interface, where the callback interface is disposed on an event path of the cache code;

The first one or two units are used to calculate the cumulative number of cache read failure events and other cache events for each data.

Further, the preset classification criteria include classification according to a function of data or a unique identifier of data.

Further, the second device is configured to generate a graph according to the total number of read failure events and other cache events in the event record of the acquired similar data, and determine a cache hit ratio according to the chart. the reason.

Compared with the prior art, the present application obtains data with a low cache hit ratio according to the total number of data cache read failure events, and the cumulative number of read failure events and other cache events according to data having a low cache hit ratio. , can effectively determine the reason for the low cache hit rate, help to improve the quality of the cache code and improve the overall system performance.

Further, the present application is based on a combination of at least one of a cumulative number of read failure events (GetMiss) of each data and a cumulative number of delete events (DeleteExist), write events (SetNotExist), and elimination events (Evict). The reason why the data cache hit rate is low can be effectively determined.

Further, the present application can accurately monitor and record events of all data through a callback interface, and subsequently obtain a cumulative number of cache read failure events and other cache events for each data according to the event of the recorded data.

Further, the present application generates a graph according to the total number of read failure events and other cache events in the event record of the same type of data obtained, which can be generated according to the generated graph. More intuitively, the reason for the low hit rate.

DRAWINGS

Other features, objects, and advantages of the present application will become more apparent from the detailed description of the accompanying drawings.

1 shows a flow chart of a method of cache hit ratio analysis in accordance with an aspect of the present application;

2 is a flow chart showing a method of cache hit ratio analysis according to a preferred embodiment of the present application;

3 is a flow chart showing a method of cache hit ratio analysis of another preferred embodiment of the present application;

4 is a flow chart showing a method of cache hit ratio analysis in still another preferred embodiment of the present application;

FIG. 5 is a color diagram showing the cumulative number of all data read failure events and other cache events in an embodiment of the present application; FIG.

6 shows a block diagram of a device for cache hit ratio analysis in accordance with another aspect of the present application;

7 shows a structural diagram of a device for cache hit ratio analysis in accordance with a preferred embodiment of the present application;

FIG. 8 is a block diagram showing an apparatus for cache hit ratio analysis according to another preferred embodiment of the present application; FIG.

The same or similar reference numerals in the drawings denote the same or similar components.

detailed description

In a typical configuration of the present application, the terminal, the device of the service network, and the trusted party each include one or more processors (CPUs), input/output interfaces, network interfaces, and memory.

The memory may include non-persistent memory, random access memory (RAM), and/or non-volatile memory in a computer readable medium, such as read only memory (ROM) or flash memory. Memory is an example of a computer readable medium.

Computer readable media includes both permanent and non-permanent, removable and non-removable media Information storage is implemented by any method or technique. The information can be computer readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read only memory. (ROM), electrically erasable programmable read only memory (EEPROM), flash memory or other memory technology, compact disk read only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, A magnetic tape cartridge, magnetic tape storage or other magnetic storage device or any other non-transportable medium that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include non-transitory computer readable media, such as modulated data signals and carrier waves.

As shown in FIG. 1 , the present application provides a method for analyzing a cache hit ratio, including:

Step S1: Obtain an event record of the same type of data in which the total number of cache read failure events is higher than the first preset threshold; wherein the read failure event (GetMiss) is generated by a Get request initiated by the user, and the read result is not If the sum of the cache read failure events is higher than the first preset threshold, the cache hit ratio of the data may be determined to be low, and the cache hit ratio of the data may be determined in a subsequent step. the reason;

In step S2, the reason why the data cache hit ratio is low is determined according to the cumulative number of read failure events and other cache events of each data in the event record of the acquired similar data. Here, the sum of the cache read failure events of a certain type of data is equal to the sum of the cumulative number of cache read failure events of all the data in the data, and the reasons for the low cache hit ratio are basically as follows. the elements of:

1. Insufficient cache capacity: When the capacity is insufficient, the cache system will be eliminated, resulting in a lower probability of data being cached. Although the cache system has the LRU algorithm to ensure that the phase-out data is non-hotspot data, it cannot guarantee the stability of the hit rate in essence. Here, the elimination (evict) means that the general cache system has a limited capacity. When the storage capacity is insufficient, the cache system will be eliminated, that is, some data is deleted to release the space. The strategy for deleting data is generally calculated by LRU (Least Recently Used). law. Among them, the LRU algorithm has the least recently used algorithm, which is a data structure algorithm, which can find the least recently used element, which is generally used in the Cache elimination strategy, that is, the least recently used element is deleted to achieve the release space. the goal of.

2. Data is actively deleted: According to the typical usage scenario of the cache system, when the data is modified or deleted, the corresponding data in the cache needs to be deleted, and the next access will inevitably occur.

3. The accessed data does not exist: the data does not exist in the entire business system and must not be hit. Each of the data can be searched according to the corresponding Key, and Key is a unique index of the data in the cache, and is not repeated in the same cache system.

4. The accessed data does not have hotspots: according to the usage mode of the cache system, when the first access discovery data is not hit, the data needs to be read from the backend data source to be written into the cache system, and the second access is hit. Assume that in extreme cases, each data will only be accessed once, and the theoretical hit rate is zero. The present application obtains data with a low cache hit ratio according to the total number of data cache read failure events, and can effectively determine the cache hit ratio according to the number of read failure events and other cache events with low cache hit ratio. The low reason helps to improve the quality of the cache code and improve the overall system performance.

In a preferred embodiment of the cache hit ratio analysis of the present application, the other cache events include one or any combination of a delete event (DeleteExist), a write event (SetNotExist), and an eliminator event (Evict). The delete event (DeleteExist) is generated by a user-initiated Delete request, and the data specified in the delete event exists before the deletion; the write event (SetNotExist) is generated by the user-initiated Set request, and is written in the event. The specified data does not exist before deletion; the Evict event refers to the elimination action when the cache capacity is full. The data can be effectively determined according to the combination of the cumulative number of read failure events (GetMiss) of each data and at least one of the cumulative number of delete events (DeleteExist), write events (SetNotExist), and elimination events (Evict). The reason for the low cache hit rate. Those skilled in the art should be able to understand that the descriptions of the other cache events described above are merely examples, and other descriptions of other existing or future cache events may be applicable to the present application, and should also be included in the protection of the present application. It is within the scope and is hereby incorporated by reference.

In a preferred embodiment of the cache hit ratio analysis of the present application, in step S2, the data cache hit ratio is determined according to the cumulative number of read failure events and other cache events of each data in the event record of the acquired homogeneous data. Reasons for low, including:

When the cumulative number of delete events (deleteExistCount) and read failure events (getMissCount) of a certain data are both greater than the second preset threshold, and the cumulative number of delete events (deleteExistCount) and the read failure event If the difference of the accumulated quantity (getMissCount) is less than the third preset threshold, it is determined that the data cache hit rate is low because the data is actively deleted. For example, as shown in Table 1, for the data record whose Key is MainPage, according to the name of the MainPage, it can be determined that the Key content is the first page data, and the same type of data of the data has only one, that is, itself, but its single read failure event. The cumulative number (getMissCount) is 30, which means that it is a low hit rate data, and the cumulative number of delete events (deleteExistCount) is also higher than 30, indicating that the key data is modified too frequently, resulting in a hit rate. low. Those skilled in the art should understand that the above description of determining the reason why the data cache hit rate is low is only an example, and other existing or future possible reasons for determining the data cache hit rate are as applicable to the present application, and should also be included in The scope of the present application is intended to be included herein by reference.

-{Key:MainPage,deleteExistCount:30,evictCount:0, getMissCount:30,setNotExistCount:30}
-{Key:item_09cdec2d,deleteExistCount:0,evictCount:11, getMissCount:8,setNotExistCount:8}
-{Key:item_122cf833,deleteExistCount:1,evictCount:14, getMissCount:9,setNotExistCount:9}
-{Key:item_14a3a03a,deleteExistCount:0,evictCount:14, getMissCount:9,setNotExistCount:9}
-{Key:item_15ecb7bd,deleteExistCount:0,evictCount:14, getMissCount:9,setNotExistCount:9}
-{Key:item_17085737,deleteExistCount:0,evictCount:14, getMissCount:9,setNotExistCount:9}
-{Key:item_1c902fdd,deleteExistCount:2,evictCount:13, getMissCount:8,setNotExistCount:8}
-{Key:item_1ef6c4bc,deleteExistCount:2,evictCount:13, getMissCount:8,setNotExistCount:8}
-{Key:item_1fd78837,deleteExistCount:1,evictCount:12, getMissCount:9,setNotExistCount:9}

-{Key:item_1ff9b528,deleteExistCount:0,evictCount:14, getMissCount:9,setNotExistCount:9}
-{Key:item_31c3e0a6,deleteExistCount:0,evictCount:12, getMissCount:9,setNotExistCount:9}
-{Key:item_3752aa0a,deleteExistCount:0,evictCount:11, getMissCount:8,setNotExistCount:8}
-{Key:item_3c77cac2,deleteExistCount:1,evictCount:13, getMissCount:8,setNotExistCount:8}
-{Key:item_43eaf434,deleteExistCount:0,evictCount:11, getMissCount:8,setNotExistCount:8}
-{Key:item_4511044f,deleteExistCount:0,evictCount:13, getMissCount:8,setNotExistCount:8}
-{Key:item_4f8f3999,deleteExistCount:0,evictCount:14, getMissCount:9,setNotExistCount:9}
-{Key:item_52fa54d4,deleteExistCount:2,evictCount:12, getMissCount:9,setNotExistCount:9}
-{Key:item_55fd3295,deleteExistCount:1,evictCount:12, getMissCount:9,setNotExistCount:9}
-{Key:item_58ea547f,deleteExistCount:1,evictCount:14, getMissCount:9,setNotExistCount:9}
-{Key:item_5afffbfe,deleteExistCount:2,evictCount:14, getMissCount:9,setNotExistCount:9}
-{Key:item_67bf2f27,deleteExistCount:0,evictCount:11, getMissCount:8,setNotExistCount:8}
-{Key:item_69c42d6e,deleteExistCount:0,evictCount:12, getMissCount:9,setNotExistCount:9}
-{Key:item_6bb5a7da,deleteExistCount:2,evictCount:14, getMissCount:9,setNotExistCount:9}
-{Key:item_7466fb35,deleteExistCount:2,evictCount:13, getMissCount:8,setNotExistCount:8}
-{Key:item_7b59ef31,deleteExistCount:2,evictCount:11, getMissCount:8,setNotExistCount:8}
-{Key:item_7e26abc2,deleteExistCount:1,evictCount:11, getMissCount:8,setNotExistCount:8}
-{Key:item_82a5b1a8,deleteExistCount:0,evictCount:12, getMissCount:9,setNotExistCount:9}
-{Key:item_842c6a4a,deleteExistCount:1,evictCount:12, getMissCount:9,setNotExistCount:9}
-{Key:item_880242f6,deleteExistCount:2,evictCount:11, getMissCount:8,setNotExistCount:8}
-{Key:item_8d446f69,deleteExistCount:0,evictCount:14, getMissCount:9,setNotExistCount:9}
-{Key:item_8df56967,deleteExistCount:0,evictCount:13, getMissCount:8,setNotExistCount:8}
-{Key:item_8ebebdfd,deleteExistCount:0,evictCount:13, getMissCount:8,setNotExistCount:8}
-{Key:item_91096a22,deleteExistCount:2,evictCount:11, getMissCount:8,setNotExistCount:8}
-{Key:item_992b919c,deleteExistCount:0,evictCount:13, getMissCount:8,setNotExistCount:8}
-{Key:item_a4f1bb1e,deleteExistCount:1,evictCount:13, getMissCount:8,setNotExistCount:8}
-{Key:item_aa8892f5,deleteExistCount:0,evictCount:13, getMissCount:8,setNotExistCount:8}
-{Key:item_ac4985e3,deleteExistCount:0,evictCount:13, getMissCount:8,setNotExistCount:8}
-{Key:item_b57baa59,deleteExistCount:0,evictCount:12, getMissCount:9,setNotExistCount:9}
-{Key:item_be813651,deleteExistCount:2,evictCount:12, getMissCount:9,setNotExistCount:9}
-{Key:item_c1edea98,deleteExistCount:1,evictCount:11, getMissCount:8,setNotExistCount:8}
-{Key:item_c254be5b,deleteExistCount:0,evictCount:12, getMissCount:9,setNotExistCount:9}
-{Key:item_c4385493,deleteExistCount:2,evictCount:12, getMissCount:9,setNotExistCount:9}
-{Key:item_c86d52f4,deleteExistCount:0,evictCount:12, getMissCount:9,setNotExistCount:9}
-{Key:item_cf726c32,deleteExistCount:0,evictCount:11, getMissCount:8,setNotExistCount:8}
-{Key:item_e8b98eab,deleteExistCount:1,evictCount:13, getMissCount:8,setNotExistCount:8}

-{Key:item_e92c7bfc,deleteExistCount:1,evictCount:14, getMissCount:9,setNotExistCount:9}
-{Key:item_eda48d55,deleteExistCount:0,evictCount:11, getMissCount:8,setNotExistCount:8}
-{Key:item_f05b915a,deleteExistCount:0,evictCount:11, getMissCount:8,setNotExistCount:8}
-{Key:item_f220ecd9,deleteExistCount:1,evictCount:11, getMissCount:8,setNotExistCount:8}
-{Key:item_f6087d9d,deleteExistCount:2,evictCount:14, getMissCount:9,setNotExistCount:9}
-{Key:misc_foot,setdeleteExistCount:0,evictCount:0, getMissCount:20,NotExistCount:0}
-{Key:nick_0282d81c,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:nick_149a3e16,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:nick_1c8768b5,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:nick_280f3dd2,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:nick_5bbecebd,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:nick_640aaa5e,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:nick_66608a9f,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:nick_684b984a,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:nick_6da33be3,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:nick_71b6db52,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:nick_72341f21,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:nick_9a121e06,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:nick_9aee2b6f,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:nick_a592aedc,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:nick_a8608f23,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:nick_a91e57aa,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:nick_a9e261db,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:nick_b0b76c12,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:nick_b19ae545,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:nick_b44dede3,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:nick_bd335833,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:nick_d3273f07,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:nick_dd25603e,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:nick_eadeaead,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:nick_f21f0ce4,deleteExistCount:0,victCount:0, getMissCount:1,setNotExistCount:1}
-{Key:session_0197f1f9,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:session_0f10bac7,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:session_104ba3c7,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:session_146af489,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:session_1fb8dbd7,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}

-{Key:session_2410cd60,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:session_284dfa9a,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:session_3781cdf6,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:session_50e0ba1e,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:session_8ed5c998,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:session_a3dc28b2,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:session_b9a9e2d2, deleteExistCount: 0, evictCount: 0, getMissCount: 1, setNotExistCount: 1}
-{Key:session_c7936ceb,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:session_ff83d040,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:1}
-{Key:token_3a492fbc,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:0}
-{Key:token_4f58ca30,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:0}
-{Key:token_58f50b6f,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:0}-{Key:token_5e1cb9a6,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:0}
-{Key:token_6824fe45,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:0}
-{Key:token_84fd1778,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:0}
-{Key:token_ae211626,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:0}
-{Key:token_be5e931f,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:0}
-{Key:token_d4529ce6,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:0}
-{Key:token_f6db4f04,deleteExistCount:0,evictCount:0, getMissCount:1,setNotExistCount:0}

Table 1

In a preferred embodiment of the cache hit ratio analysis of the present application, in step S2, the data cache hit ratio is determined according to the cumulative number of read failure events and other cache events of each data in the event record of the acquired homogeneous data. Reasons for low, including:

When the cumulative number of read failure events of a certain data (getMissCount) is greater than the fourth preset threshold, and the difference between the cumulative number of read failure events (getMissCount) and the cumulative number of write events (setNotExistCount) is greater than the fifth pre- If the threshold is set, it is determined that the data cache hit rate is low because the accessed data does not exist. For example, for the data in Table 1 whose key is misc_foot, or a set of event data of the same type with Key as the token_ prefix, here, the data of the key for misc_foot has only one type of data, namely itself, but it is read. The cumulative number of failure events (getMissCount) is 20, indicating that the cache hit rate of the data is low, and the cumulative number of read failure events (getMissCount) for each data in the same type of data whose Key is the token_ prefix is 1, but The total number of times of the cumulative number of read failure events (getMissCount) of the same data with the key of the token_ prefix is higher, that is, the number of data in the same type of data is larger, indicating that the same type of data cache hit with Key as the token_ prefix The rate is low. According to the detailed information of each data record, the value of the write event (SetNotExistCount) is 0, indicating that the service system does not cache the data whose key is misc_foot or the data whose key is token_ prefix, resulting in low hit rate. Those skilled in the art should understand that the above description of determining the reason why the data cache hit rate is low is only an example, and other existing or future possible reasons for determining the data cache hit rate are as applicable to the present application, and should also be included in The scope of the present application is intended to be included herein by reference.

In a preferred embodiment of the cache hit ratio analysis of the present application, in step S2, the data cache hit ratio is determined according to the cumulative number of read failure events and other cache events of each data in the event record of the acquired homogeneous data. Reasons for low, including:

When the cumulative number of elimination events (evictCount) of a certain data and the cumulative number of read failure events (getMissCount) are both greater than the sixth preset threshold, and the cumulative number of eliminated events (evictCount) and the cumulative number of read failure events ( If the difference of getMissCount) is less than the seventh preset threshold, it is determined that the data cache hit rate is low because the cache capacity is insufficient. For example, in Table 1, the total number of times of the total number of read failure events (getMissCount) of the same type of data prefixed by item is relatively high, that is, the number of data in the same type of data is large, and it is found out based on the detailed information thereof. The cumulative number of events (evictCount) is also high, indicating the elimination of the lack of cache capacity, resulting in a drop in hit rate. Those skilled in the art should understand that the above description of determining the reason why the data cache hit rate is low is only an example, and other existing or future possible reasons for determining the data cache hit rate are as applicable to the present application, and should also be included in The scope of the present application is intended to be included herein by reference.

In an optional embodiment of the cache hit ratio analysis of the present application, the second threshold, the fourth threshold, and the sixth threshold may be set to the same value.

Method for Cache Hit Ratio Analysis of the Present Application In a preferred embodiment, step S2 is obtained according to The cumulative number of read failure events and other cache events for each data in the event record of the same type of data obtained determines the reason why the data cache hit rate is low, including:

When the cumulative number of write events (setNotExistCount) of a certain data and the cumulative number of read failure events (getMissCount) are both smaller than the eighth preset threshold, and the cumulative number of write events (setNotExistCount) and the read failure If the accumulated number of events (getMissCount) is less than the ninth preset threshold, it is determined that the data cache hit rate is low because the accessed data does not have hotspots. For example, the cumulative number of read failure events (GetMissCount) of the same type of data prefixed by session in Table 1 is basically 1 or 2, but the total number of totals of the read failure events (GetMissCount) is larger, that is, the same type of data. The number of data in the data is large, and the write event (SetNotExistCount) of each data record is basically 1, indicating that the hotspot of this type of data is not high. Those skilled in the art should understand that the above description of determining the reason why the data cache hit rate is low is only an example, and other existing or future possible reasons for determining the data cache hit rate are as applicable to the present application, and should also be included in The scope of the present application is intended to be included herein by reference.

As shown in FIG. 2, in a preferred embodiment of the cache hit ratio analysis of the present application, in step S1, an event record of the same type of data whose sum of cache read failure events is higher than a first preset threshold is obtained, including:

In step S11, the cumulative number of cache read failure events and other cache events of each data is obtained. As shown in FIG. 3, in a preferred embodiment of the cache hit ratio analysis method of the present application, step S11 includes step S111. And step S112, wherein, in step S111, a cache read failure event or other cache event of each data is obtained through a callback interface, where the callback interface is set on an event path of the cache code; here, each data can be returned through the interface The cache read failure event or other cache event is monitored, and the information of the monitored event may be recorded one by one according to the sequence of occurrence of the event, including what kind of event is such as a read failure event or other cache event, and the event corresponding What is the key of the data, the recording method can be implemented by file appending and writing, and an event log file can be generated after a specified period of time, due to the callback interface. Set on the event path of the cache code, the callback interface can monitor and record the event whenever an event occurs; and in step S112, calculate the cumulative number of cache read failure events and other cache events for each data. Here, the cumulative number of cache read failure events (GetMissCount) and the cumulative number of other cache events (EvictCount, DeleteExistCount, SetNotExistCount) of each data may be calculated according to the content of the event record file generated in the above step S111. For example, the event record file can be read in order, and the corresponding Key is recorded in the memory (or other storage medium), and the number of event records of the key is accumulated. After the record scan is completed, a [Key: EvictCount, DeleteExistCount, The list of Keys of GetMissCount, SetNotExistCount], those skilled in the art should be able to understand that the above description of the cumulative number of cache read failure events and other cache events for each data is only an example, other existing or future acquisitions may occur. A description of the cumulative number of cached read failure events and other cached events of the data, as applicable to the present application, is also intended to be included within the scope of the present application and is hereby incorporated by reference herein;

In step S12, the data that the cumulative number of read failure events is greater than the tenth preset threshold is filtered out from all the data; since the size of the Key list generated by all the data is relatively large, which is not conducive to the efficiency of subsequent analysis, The cumulative number of read failure events (GetMissCount) in the event record of each data, from the largest to the smallest N data before the screening is greater than the tenth preset threshold, where N is too small will affect the accuracy of the data, N too The assembly affects the view display, so you can set N to the default value according to the size of the data, such as 100, that is, select an event record of up to 100 data;

Step S13, each of the filtered data is classified into a corresponding class according to a preset classification criterion;

Step S14, calculating the total number of cache read failure events of each type of data;

Step S15: Obtain an event record of the same type of data whose sum of cache read failure events is higher than a first preset threshold, where the event record of the same type data includes a read failure event of each data in the type of data and other The cumulative number of cached events. Here, if the cache is read out If the sum of the failure events is higher than the first preset threshold, it may be determined that the cache hit ratio of the data is low, and the reason why the cache hit ratio of the data is low may be determined in the subsequent step. There are three cases in which the total number of cache read failure events is higher than the first preset threshold. In one case, the number of event records of data in a certain type of data is small, but the cache read failure event of each data is The cumulative number is high, and the total number of cache read failure events after the sum of these accumulated numbers is higher than the first preset threshold; the second case is that although the cache read failure event of each data in a certain type of data The cumulative number is small, but the number of event records of the data in the data is large, and the total number of cache read failure events after the cumulative number is added is still higher than the first preset threshold; the third case is In the first and second cases, the cumulative number of cache read failure events for a single data is high, and the cumulative number of cache read failure events for multiple data is added. This is a very high case. Finally, the sum of the cache read failure events after the sum of these cumulative quantities is still higher than the first preset threshold. Those skilled in the art should be able to understand that the description of the event record of the same type of data in which the sum of the cache read failure events is higher than the first preset threshold is only an example, and other existing or future acquisition cache read failures may occur. The description of the event record of the same type of data in which the sum of the events is higher than the first predetermined threshold is applicable to the present application and is also included in the scope of the present application and is hereby incorporated by reference.

In a preferred embodiment of the cache hit ratio analysis of the present application, the preset classification criteria in step S13 includes classifying by function of data or by unique identifier (Key) of data. Keys in the cache system generally have prefix consistency, which means that the same type of function will be expressed by the same prefix to distinguish: for example, in the Key of Table 1, the same type of service with the same prefix, such as item, session, nick Equal prefixes indicate different classes. A person skilled in the art should understand that the description of the foregoing preset classification standard is only an example, and other existing or future possible preset classification standards may be applicable to the present application and should also be included in the scope of protection of the present application. It is hereby incorporated by reference.

As shown in FIG. 4, in a preferred embodiment of the method for cache hit ratio analysis of the present application, Step S2, determining, according to the cumulative number of read failure events and other cache events of each data in the event record of the acquired similar data, the reason why the data cache hit rate is low, including:

Step S21, generating all the data in the event record of the acquired similar data according to the size of the cumulative number of read failure events and other cache events, such as any one of a colored map, a graph, or a histogram; for example, As shown in FIG. 5, when generating a colored map, Keys may be arranged according to the lexicographic order of the Key of the data, 10 lines per line, a total of 10 lines, and each Key correspondingly draws a color block, and the background color depth of the color block is determined by the The value of the total number of read failure events (GetMissCount) of the data corresponding to the Key is determined; the maximum value of GetMissCount is defined as positive red (R255, G0, B0), and the color of GetMissCount is 0 is defined as positive green (R0, G255, B0) ), the middle value between positive red and positive green is yellow, the larger the GetMissCount, the more reddish the color, the formula is:

Assume that the largest GetMissCount in the entire series is MAX

If GetMIssCount>=MAX/2, then

R=255; G=255*2*(1-GetMissCount/MAX); B=0;

If GetMIssCount<MAX/2, then

R=255*(2*GetMissCount/MAX); G=255; B=0;

In addition, it is required to mark the cumulative number of other buffer events corresponding to the data of the color block on each color block, and the labeling manner may be to first hide the accumulated number of other buffer events on each color block, and then display the color block when the color block is selected. The cumulative number of hidden other cached events;

Step S22, determining a reason why the cache hit ratio is low according to the chart. Here, the reason why the hit rate is low can be more intuitively obtained from the generated chart. Specifically, the Key in the cache system generally has prefix consistency, that is, the same type of function is expressed by the same prefix to distinguish: for example, session_312312, session_888544, item_box111, item_box222, therefore, as shown in FIG. 5, according to Key's lexicographic order arranges the data and condenses the same business function, that is, the same type of data color blocks. According to the principle that the larger the GetMissCount is, the more reddish the color is, or the larger the arrangement area of the color blocks of the same type of data, the color with a low hit rate is determined. The block, then for the color block with a low hit rate, can easily find the reason for the low hit rate based on the cumulative number of other cache events on the color block corresponding to the data corresponding to the color block. Those skilled in the art should understand that the above description of determining the reason why the data cache hit rate is low is only an example, and other existing or future possible reasons for determining the data cache hit rate are as applicable to the present application, and should also be included in The scope of the present application is intended to be included herein by reference.

As shown in FIG. 6, according to another aspect of the present application, the present application further provides an apparatus 100 for cache hit ratio analysis, which includes:

The first device 1 is configured to obtain an event record of the same type of data whose cache read failure event is higher than the first preset threshold; wherein the read failure event (GetMiss) is generated by the user-initiated Get request, and is read. The result is non-existent. If the total number of cache read failure events is higher than the first preset threshold, it may be determined that the cache hit rate of the data is low, and then the cache of the data may be determined in a subsequent step. The reason for the low hit rate;

The second device 2 is configured to determine, according to the cumulative number of read failure events and other cache events of each data in the event record of the acquired similar data, the reason why the data cache hit rate is low. Here, the sum of the cache read failure events of a certain type of data is equal to the sum of the cumulative number of cache read failure events of all the data in the data, and the reasons for the low cache hit ratio are basically as follows. the elements of:

1. Insufficient cache capacity: When the capacity is insufficient, the cache system will be eliminated, resulting in a lower probability of data being cached. Although the cache system has the LRU algorithm to ensure that the phase-out data is non-hotspot data, it cannot guarantee the stability of the hit rate in essence. Here, the elimination (evict) means that the general cache system has a limited capacity. When the storage capacity is insufficient, the cache system will be eliminated, that is, some data is deleted to release the space. The strategy for deleting data generally adopts the LRU (Least Recently Used) algorithm. . Among them, the LRU algorithm has the least recently used algorithm, which is a data structure algorithm, which can find the least recently used element, which is generally used in the Cache elimination strategy, that is, the least recently used element is deleted to achieve the release space. the goal of.

2. Data is actively deleted: according to the typical usage scenario of the cache system, the data is modified or deleted. In addition, the corresponding data in the cache needs to be deleted, and the next access will inevitably occur.

3. The accessed data does not exist: the data does not exist in the entire business system and must not be hit. Each of the data can be searched according to the corresponding Key, and Key is a unique index of the data in the cache, and is not repeated in the same cache system.

4. The accessed data does not have hotspots: according to the usage mode of the cache system, when the first access discovery data is not hit, the data needs to be read from the backend data source to be written into the cache system, and the second access is hit. Assume that in extreme cases, each data will only be accessed once, and the theoretical hit rate is zero. The present application obtains data with a low cache hit ratio according to the total number of data cache read failure events, and can effectively determine the cache hit ratio according to the number of read failure events and other cache events with low cache hit ratio. The low reason helps to improve the quality of the cache code and improve the overall system performance.

In a preferred embodiment of the apparatus for cache hit ratio analysis of the present application, the other cache events include one or any combination of a delete event (DeleteExist), a write event (SetNotExist), and a knockout event (Evict). The delete event (DeleteExist) is generated by a user-initiated Delete request, and the data specified in the delete event exists before the deletion; the write event (SetNotExist) is generated by the user-initiated Set request, and is written in the event. The specified data does not exist before deletion; the Evict event refers to the elimination action when the cache capacity is full. The data can be effectively determined according to the combination of the cumulative number of read failure events (GetMiss) of each data and at least one of the cumulative number of delete events (DeleteExist), write events (SetNotExist), and elimination events (Evict). The reason for the low cache hit rate. Those skilled in the art should understand that the descriptions of the other cache events mentioned above are only examples, and other descriptions of other existing or future cache events may be applicable to the present application, and should also be included in the scope of the present application. This is hereby incorporated by reference.

In a preferred embodiment of the apparatus for cache hit ratio analysis, the second apparatus is configured to use a cumulative number of deletion events (deleteExistCount) and a cumulative number of read failure events (getMissCount) when a certain data is deleted. Both are greater than the second preset threshold, and the deletion of the event is tired. If the difference between the cumulative number of counts (deleteExistCount) and the cumulative number of read failure events (getMissCount) is less than the third preset threshold, it is determined that the data cache hit rate is low because the data is actively deleted. For example, as shown in Table 1, for the data record whose Key is MainPage, according to the name of the MainPage, it can be determined that the Key content is the first page data, and the same type of data of the data has only one, that is, itself, but its single read failure event. The cumulative number (getMissCount) is 30, which means that it is a low hit rate data, and the cumulative number of delete events (deleteExistCount) is also higher than 30, indicating that the key data is modified too frequently, resulting in a hit rate. low. Those skilled in the art should understand that the above description of determining the reason why the data cache hit rate is low is only an example, and other existing or future possible reasons for determining the data cache hit rate are as applicable to the present application, and should also be included in The scope of the present application is intended to be included herein by reference.

In a preferred embodiment of the device for cache hit ratio analysis, the second device is configured to: when the cumulative number of read failure events of a certain data (getMissCount) is greater than a fourth preset threshold, and read If the difference between the cumulative number of failed events (getMissCount) and the accumulated number of write events (setNotExistCount) is greater than the fifth predetermined threshold, it is determined that the data cache hit rate is low because the accessed data does not exist. For example, for the data in Table 1 whose key is misc_foot, or a set of event data of the same type with Key as the token_ prefix, here, the data of the key for misc_foot has only one type of data, namely itself, but it is read. The cumulative number of failure events (getMissCount) is 20, indicating that the cache hit rate of the data is low, and the cumulative number of read failure events (getMissCount) for each data in the same type of data whose Key is the token_ prefix is 1, but The total number of times of the cumulative number of read failure events (getMissCount) of the same data with the key of the token_ prefix is higher, that is, the number of data in the same type of data is larger, indicating that the same type of data cache hit with Key as the token_ prefix The rate is low. According to the detailed information of each data record, the value of the write event (SetNotExistCount) is 0, indicating that the service system does not cache the data whose key is misc_foot or the data whose key is token_ prefix, resulting in low hit rate. Those skilled in the art should be able to understand that the above description of determining the reason why the data cache hit rate is low is only For example, other existing or future possible determinations of a low data cache hit rate are applicable to the present application and are also included in the scope of the present application and are hereby incorporated by reference.

In a preferred embodiment of the apparatus for cache hit ratio analysis of the present application, the second means is used for the cumulative number of evict counts of a certain data (evictCount) and the cumulative number of read failure events (getMissCount) If the difference between the cumulative number of elimination events (evictCount) and the cumulative number of read failure events (getMissCount) is less than the seventh preset threshold, the reason for determining that the data cache hit rate is low is cache. Insufficient capacity. For example, in Table 1, the total number of times of the total number of read failure events (getMissCount) of the same type of data prefixed by item is relatively high, that is, the number of data in the same type of data is large, and it is found out based on the detailed information thereof. The cumulative number of events (evictCount) is also high, indicating the elimination of the lack of cache capacity, resulting in a drop in hit rate. Those skilled in the art should understand that the above description of determining the reason why the data cache hit rate is low is only an example, and other existing or future possible reasons for determining the data cache hit rate are as applicable to the present application, and should also be included in The scope of the present application is intended to be included herein by reference.

In an optional embodiment of the apparatus for cache hit ratio analysis, the second threshold, the fourth threshold, and the sixth threshold may be set to the same value.

In a preferred embodiment of the apparatus for cache hit ratio analysis of the present application, the second means is for a cumulative number of write events of a certain data (setNotExistCount) and a cumulative number of read failure events (getMissCount) ) is less than the eighth preset threshold, and the difference between the accumulated number of write events (setNotExistCount) and the cumulative number of read failure events (getMissCount) is less than the ninth preset threshold, and the reason for determining that the data cache hit ratio is low is The data accessed is not hot. For example, the cumulative number of read failure events (GetMissCount) of the same type of data prefixed by session in Table 1 is basically 1 or 2, but the total number of totals of the read failure events (GetMissCount) is larger, that is, the same type of data. The number of data in the data is large, and the write event (SetNotExistCount) of each data record is basically 1, indicating The hotspots of this type of data are not high. Those skilled in the art should understand that the above description of determining the reason why the data cache hit rate is low is only an example, and other existing or future possible reasons for determining the data cache hit rate are as applicable to the present application, and should also be included in The scope of the present application is intended to be included herein by reference.

As shown in FIG. 7, in a preferred embodiment of the device for cache hit ratio analysis, the first device 1 includes:

The first module 11 is configured to obtain a cumulative number of cache read failure events and other cache events for each data;

The first two modules 12 are configured to filter data from all the data that the cumulative number of read failure events is greater than the tenth preset threshold; the size of the Key list generated by all the data is relatively large, which is not conducive to subsequent analysis. Efficiency, so according to the total number of read failure events (GetMissCount) in the event record of each data, the top N data is larger than the tenth preset threshold from the largest to the smallest, where N is too small will affect the data. Accuracy, N is too large to affect the view display, so you can set N to the default value according to the size of the data, such as 100, that is, select an event record of up to 100 data;

The first three modules 13 are configured to classify each of the filtered data into corresponding classes according to a preset classification criterion;

The first four module 14 is configured to calculate a total number of cache read failure events of each type of data;

The first five module 15 is configured to obtain an event record of the same type of data in which the sum of the cache read failure events is higher than the first preset threshold, wherein the event record of the same type of data includes reading of each data in the type of data. Take the cumulative number of failed events and other cached events. Here, if the total number of cache read failure events is higher than the first preset threshold, it may be determined that the cache hit ratio of the data is low, and the reason why the cache hit ratio of the data is low may be determined in the subsequent step. There are three cases in which the total number of cache read failure events is higher than the first preset threshold. In one case, the number of event records of data in a certain type of data is small, but the cache read of each data is lost. The cumulative number of defeat events is high, and the total number of cache read failure events after the sum of these cumulative numbers is higher than the first preset threshold; the second case is that although each data in a certain type of data is cached The cumulative number of failed events is small, but the number of event records of the data in the data is large, and the total number of cache read failure events after the cumulative number is added is still higher than the first preset threshold; In the case of the first and second cases, the cumulative number of cache read failure events for a single data is high, and the cumulative number of cache read failure events for multiple data is added. In the high case, the total number of cache read failure events that are eventually added to these cumulative quantities is still higher than the first preset threshold. Those skilled in the art should be able to understand that the description of the event record of the same type of data in which the sum of the cache read failure events is higher than the first preset threshold is only an example, and other existing or future acquisition cache read failures may occur. The description of the event record of the same type of data in which the sum of the events is higher than the first predetermined threshold is applicable to the present application and is also included in the scope of the present application and is hereby incorporated by reference.

As shown in FIG. 8, in a preferred embodiment of the device for cache hit ratio analysis, the first module 11 includes:

The first one-to-one unit 111 is configured to obtain a cache read failure event or other cache event of each data through a callback interface, where the callback interface is set on an event path of the cache code; here, each data can be returned through the interface The cache read failure event or other cache event is monitored, and the information of the monitored event may be recorded one by one according to the sequence of occurrence of the event, including what kind of event is such as a read failure event or other cache event, and the event corresponding What is the key of the data, the recording method can be implemented by file append and write mode, and an event log file can be generated after a specified period of time. Since the callback interface is set in the event path of the cache code, as long as an event occurs, the callback interface is Can listen to and record the event;

The first one-two unit 112 is configured to calculate a cumulative number of cache read failure events and other cache events for each data. Here, the accumulation of the cache read failure event (GetMissCount) of each data can be calculated according to the content summary of the event record file generated by the first one-unit 111 described above. The cumulative number of quantities and other cached events (EvictCount, DeleteExistCount, SetNotExistCount). For example, the event record file can be read in order, and the corresponding Key is recorded in the memory (or other storage medium), and the number of event records of the key is accumulated. After the record scan is completed, a [Key: EvictCount, DeleteExistCount, The list of Keys of GetMissCount, SetNotExistCount], those skilled in the art should be able to understand that the above description of the cumulative number of cache read failure events and other cache events for each data is only an example, other existing or future acquisitions may occur. A description of the cumulative number of cached read failure events and other cached events of the data, as applicable to the present application, is also intended to be included within the scope of the present application and is incorporated herein by reference.

In a preferred embodiment of the apparatus for cache hit ratio analysis of the present application, the preset classification criteria include classification according to a function of data or a unique identifier (Key) of data. Keys in the cache system generally have prefix consistency, which means that the same type of function will be expressed by the same prefix to distinguish: for example, in the Key of Table 1, the same type of service with the same prefix, such as item, session, nick Equal prefixes indicate different classes. A person skilled in the art should understand that the description of the foregoing preset classification standard is only an example, and other existing or future possible preset classification standards may be applicable to the present application and should also be included in the scope of protection of the present application. It is hereby incorporated by reference.

In a preferred embodiment of the apparatus for cache hit ratio analysis, the second apparatus is configured to press all the data in the event record of the acquired similar data according to the cumulative number of read failure events and other cache events. The size of the generated chart, based on the chart to determine the reason for the low cache hit rate. The chart is any one of a colored graph, a graph, or a histogram; for example, as shown in FIG. 5, when generating a colored graph, Keys may be arranged according to the lexicographic order of the Key of the data, 10 per line, for a total of 10 Row, each key correspondingly draws a color block, the background color depth of the color block is determined by the value of the cumulative number of read failure events (GetMissCount) of the data corresponding to the Key; the maximum value of GetMissCount is defined as positive red (R255, G0, B0), GetMissCount is 0 The color is defined as positive green (R0, G255, B0). The middle value between positive red and positive green is yellow. The larger the GetMissCount is, the more reddish the color is. The formula is:

Assume that the largest GetMissCount in the entire series is MAX

If GetMIssCount>=MAX/2, then

R=255; G=255*2*(1-GetMissCount/MAX); B=0;

If GetMIssCount<MAX/2, then

R=255*(2*GetMissCount/MAX); G=255; B=0;

In addition, it is required to mark the cumulative number of other buffer events corresponding to the data of the color block on each color block, and the labeling manner may be to first hide the accumulated number of other buffer events on each color block, and then display the color block when the color block is selected. The cumulative number of hidden other cached events;

Subsequently, the reason for the low hit rate can be obtained more intuitively based on the generated chart. Specifically, the Key in the cache system generally has prefix consistency, that is, the same type of function is expressed by the same prefix to distinguish: for example, session_312312, session_888544, item_box111, item_box222, therefore, as shown in FIG. 5, according to Key's lexicographic order arranges the data and condenses the same business function, that is, the same type of data color blocks. According to the principle that the larger the GetMissCount is, the more reddish the color is, or the larger the arrangement area of the color blocks of the same type of data, the color block with a low hit rate is determined, and then the color block with a low hit rate is determined according to each color. The cumulative number of other cache events on the block labeled with the color block can easily find the reason for the low hit rate. Those skilled in the art should understand that the above description of determining the reason why the data cache hit rate is low is only an example, and other existing or future possible reasons for determining the data cache hit rate are as applicable to the present application, and should also be included in The scope of the present application is intended to be included herein by reference.

In summary, the present application obtains data with a low cache hit ratio according to the total number of data cache read failure events, and according to the cumulative number of read failure events and other cache events with low cache hit ratio, Effectively determining the reason for the low cache hit rate helps to improve the quality of the cache code and improve overall system performance.

Further, the present application is based on the accumulation of a read failure event (GetMiss) that can be used for each data. The combination of the quantity and the cumulative number of delete events (DeleteExist), write events (SetNotExist), and elimination events (Evict) can effectively determine the reason why the data cache hit rate is low.

Further, the present application can accurately monitor and record events of all data through a callback interface, and subsequently obtain a cumulative number of cache read failure events and other cache events for each data according to the event of the recorded data.

Further, the present application generates a graph according to the total number of read failure events and other cache events in the event record of the same type data obtained by the present application, and can more intuitively obtain the reason of the low hit rate according to the generated graph.

It will be apparent to those skilled in the art that various modifications and changes can be made in the present application without departing from the spirit and scope of the application. Thus, it is intended that the present invention cover the modifications and variations of the present invention.

It should be noted that the present application can be implemented in software and/or a combination of software and hardware, for example, using an application specific integrated circuit (ASIC), a general purpose computer, or any other similar hardware device. In one embodiment, the software program of the present application can be executed by a processor to implement the steps or functions described above. Likewise, the software programs (including related data structures) of the present application can be stored in a computer readable recording medium such as a RAM memory, a magnetic or optical drive or a floppy disk and the like. In addition, some of the steps or functions of the present application may be implemented in hardware, for example, as a circuit that cooperates with a processor to perform various steps or functions.

In addition, a portion of the present application can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide a method and/or technical solution in accordance with the present application. The program instructions for invoking the method of the present application may be stored in a fixed or removable recording medium, and/or transmitted by a data stream in a broadcast or other signal bearing medium, and/or stored in a The working memory of the computer device in which the program instructions are run. Here, an embodiment according to the present application includes a device, the device package A memory for storing computer program instructions and a processor for executing program instructions, wherein when the computer program instructions are executed by the processor, the device is triggered to operate based on the foregoing methods and embodiments according to various embodiments of the present application / or technical solutions.

It is obvious to those skilled in the art that the present application is not limited to the details of the above-described exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present application. Therefore, the present embodiments are to be considered as illustrative and not restrictive, and the scope of the invention is defined by the appended claims instead All changes in the meaning and scope of equivalent elements are included in this application. Any reference signs in the claims should not be construed as limiting the claim. In addition, it is to be understood that the word "comprising" does not exclude other elements or steps. A plurality of units or devices recited in the device claims may also be implemented by a unit or device by software or hardware. The first, second, etc. words are used to denote names and do not denote any particular order.

Claims (20)

1. A method for cache hit rate analysis, which includes:

   Obtaining an event record of the same type of data whose cache read failure event is higher than the first preset threshold;

   The reason why the data cache hit rate is low is determined according to the cumulative number of read failure events and other cache events of each data in the event record of the acquired similar data.
2. The method of claim 1 wherein said other cache events comprise one or any combination of a delete event, a write event, and a retirement event.
3. The method according to claim 2, wherein the reason why the data cache hit rate is low is determined according to the cumulative number of read failure events and other cache events of each data in the event record of the acquired homogeneous data, including:

   When the cumulative number of deletion events and read failure events of a certain data is greater than a second preset threshold, and the difference between the cumulative number of deletion events and the cumulative number of read failure events is less than a third preset threshold, determining The reason for the low data cache hit rate is that the data is actively deleted.
4. The method according to claim 2 or 3, wherein the reason why the data cache hit rate is low is determined according to the cumulative number of read failure events and other cache events of each data in the event record of the acquired homogeneous data, including:

   When the accumulated number of read failure events of a certain data is greater than a fourth preset threshold, and the difference between the cumulative number of read failure events and the accumulated number of write events is greater than a fifth preset threshold, determining that the data cache hits The reason for the low rate is that the accessed data does not exist.
5. The method according to any one of claims 2 to 4, wherein the reason why the data cache hit rate is low is determined based on the cumulative number of read failure events and other cache events of each data in the event record of the acquired homogeneous data. ,include:

   When the cumulative number of the defunct event and the read failure event of a certain data is greater than the sixth preset threshold, and the difference between the cumulative number of the culling event and the accumulated number of the read failure event is less than the seventh preset threshold, determining the The reason for the low data cache hit rate is that the cache capacity is insufficient.
6. The method according to any one of claims 2 to 5, wherein the reason why the data cache hit rate is low is determined based on the cumulative number of read failure events and other cache events of each data in the event record of the acquired homogeneous data. ,include:

   When the cumulative number of the write event and the read failure event of a certain data is less than the eighth preset threshold, and the difference between the accumulated number of the write event and the accumulated number of the read failure event is less than the ninth preset threshold, then determining The reason why the data cache hit rate is low is that the accessed data is not hot.
7. The method according to any one of claims 1 to 6, wherein the event record of the same type of data in which the total number of times of the cache read failure event is higher than the first preset threshold is obtained, including:

   Get the cumulative number of cache read failure events and other cache events for each data;

   Filtering data from all the data that the cumulative number of read failure events is greater than the tenth preset threshold;

   Each of the filtered data is assigned to a corresponding class according to a preset classification criterion;

   Calculate the total number of cache read failure events for each type of data;

   Obtaining an event record of the same type of data whose sum of cache read failure events is higher than a first preset threshold, wherein the event record of the same type of data includes a read failure event of each data in the type of data and other cache events The cumulative quantity.
8. The method of claim 7 wherein obtaining a cumulative number of cache read failure events and other cache events for each of the data comprises:

   Obtaining a cache read failure event or other cache event for each data through a callback interface, the callback interface being disposed on an event path of the cache code;

   Calculates the cumulative number of cache read failure events and other cache events for each data.
9. The method of claim 7 or 8, wherein the predetermined classification criteria comprises classification by function of data or by unique identification of data.
10. The method according to any one of claims 1 to 2, wherein the reason why the data cache hit rate is low is determined based on the cumulative number of read failure events and other cache events of each data in the event record of the acquired homogeneous data. , including:

    All the data in the event record of the acquired similar data is generated according to the size of the cumulative number of read failure events and other cache events;

    The reason for the low cache hit ratio is determined based on the chart.
11. A device for cache hit ratio analysis, including:

    a first device, configured to acquire an event record of the same type of data whose cache read failure event is higher than a first preset threshold;

    And a second device, configured to determine, according to the accumulated number of read failure events and other cache events of each data in the event record of the acquired similar data, the reason why the data cache hit rate is low.
12. The device of claim 11 wherein the other cache events comprise one or any combination of a delete event, a write event, and a retirement event.
13. The device according to claim 12, wherein said second means is configured to: when the cumulative number of deletion events and read failure events of a certain data are greater than a second predetermined threshold, and the cumulative number of deletion events and reading If the difference between the cumulative number of failed events is less than the third preset threshold, it is determined that the data cache hit rate is low because the data is actively deleted.
14. The device according to claim 12 or 13, wherein the second means is configured to: when the cumulative number of read failure events of a certain data is greater than a fourth preset threshold, and the cumulative number and write of the read failure events If the difference between the accumulated number of incoming events is greater than the fifth preset threshold, it is determined that the data cache hit rate is low because the accessed data does not exist.
15. The device according to any one of claims 12 to 14, wherein the second device is configured to: when the cumulative number of the elimination events and the read failure events of a certain data are greater than a sixth preset threshold, and the elimination event The difference between the cumulative number of the cumulative number and the cumulative number of read failure events is less than the seventh predetermined threshold, and the reason for determining that the data cache hit rate is low is that the cache capacity is insufficient.
16. The device according to any one of claims 12 to 15, wherein the second means is configured to write when a cumulative number of write events and read failure events of a certain data is less than an eighth preset threshold If the difference between the cumulative number of incoming events and the cumulative number of read failed events is less than the ninth preset threshold, then the reason for determining that the data cache hit rate is low is that the accessed data does not have hotspots.
17. The apparatus of any of claims 11 to 16, wherein the first device comprises:

    The first module is configured to obtain a cumulative number of cache read failure events and other cache events for each data;

    The first two modules are configured to filter, from all data, data that the cumulative number of read failure events is greater than a tenth preset threshold;

    The first three modules are configured to classify each of the filtered data into corresponding classes according to a preset classification criterion;

    The first four modules are used to calculate the total number of cache read failure events of each type of data;

    a first fifth module, configured to obtain an event record of the same type of data whose cache read failure event is higher than a first preset threshold, wherein the event record of the same type data includes reading of each data in the type of data The cumulative number of failed events and other cached events.
18. The device of claim 17 wherein said first module comprises:

    a first one unit for obtaining a cache read failure event or other cache event of each data through a callback interface, where the callback interface is disposed on an event path of the cache code;

    The first one or two units are used to calculate the cumulative number of cache read failure events and other cache events for each data.
19. The device of claim 17 or 18, wherein the predetermined classification criteria comprises classification by function of data or by unique identification of data.
20. The device according to any one of claims 11 or 12, wherein the second means is configured to press all the data in the event record of the acquired similar data according to the cumulative number of read failure events and other cache events. Generate a chart to determine why the cache hit rate is low based on the chart.

Images