WO2019128205A1 - 实现灰度发布的方法、装置及计算节点和系统 - Google Patents
实现灰度发布的方法、装置及计算节点和系统 Download PDFInfo
- Publication number
- WO2019128205A1 WO2019128205A1 PCT/CN2018/096686 CN2018096686W WO2019128205A1 WO 2019128205 A1 WO2019128205 A1 WO 2019128205A1 CN 2018096686 W CN2018096686 W CN 2018096686W WO 2019128205 A1 WO2019128205 A1 WO 2019128205A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- grayscale
- request
- data
- user
- official
- Prior art date
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/27—Replication, distribution or synchronisation of data between databases or within a distributed database system; Distributed database system architectures therefor
- G06F16/278—Data partitioning, e.g. horizontal or vertical partitioning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2455—Query execution
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/27—Replication, distribution or synchronisation of data between databases or within a distributed database system; Distributed database system architectures therefor
Definitions
- the present application relates to, but is not limited to, a distributed data processing technology, and more particularly to a method, apparatus, and computing node and system for implementing grayscale publishing based on a distributed database.
- the grayscale release (also known as Canary Publishing) refers to a release method that can smoothly transition between black and white.
- A/B testing can be performed, that is, some users continue to use product feature A, and another user starts to use product feature B. If the user has no objection to product feature B, then step by step Expand the scope and migrate all users to product feature B.
- the grayscale release can ensure the stability of the overall system, and the problem can be found in the initial grayscale, and then adjusted to ensure its influence.
- the commonly used grayscale publishing strategy includes: applying the corresponding traffic to the original service system and the new service system according to the list of the official user and the grayscale user.
- a series of functional modules such as a configuration center will be added accordingly.
- the application needs to maintain the old and new business codes corresponding to the original business system and the new business system, as well as retain the old and new data structures and their data content.
- the system for realizing grayscale publishing is mainly implemented by the application end and the server side, and rarely involves the database. Even if the database is involved, the grayscale user and the official user use the same set of databases, and the redundancy field is added to adapt.
- the formal user needs to use the fields A, B, and C.
- the gray user needs to use the fields A, D, and E.
- this table needs to include five fields: field A, field B, field C, field D, and field E.
- field D and field E are redundant.
- field B and field C are redundant.
- grayscale application access it is necessary to restrict access to only official data, and need to mask field D and field E; while in grayscale application access, it is necessary to restrict access to only grayscale data, and need to shield field B and Field C, such processing imposes restrictions on the business; moreover, grayscale service modification becomes more and more complicated, which is not conducive to post-maintenance, and is not conducive to version rollback.
- the present application provides a method, a device, and a computing node and system for realizing grayscale publishing, which can achieve the purpose of grayscale publishing simply and effectively, and reduce the amount of business transformation.
- the present application provides a method for implementing grayscale publishing, including:
- the gray scale request is parsed according to the preset fragment configuration information, and the gray scale request is determined to be the first gray scale request for requesting the gray scale data, or the second gray scale request for requesting the gray scale data and the official data;
- the present application further provides an apparatus for implementing grayscale publishing, comprising a processor and a computer readable storage medium, wherein the computer readable storage medium stores instructions, when the instructions are executed by the processor, implementing the present A method for implementing grayscale distribution in any of the embodiments.
- the application further provides an apparatus for implementing grayscale publishing, comprising: a filtering module, a parsing module, and a forwarding module; wherein
- a filtering module configured to process the received application request, and determine that the received application request is a grayscale request
- the parsing module is configured to parse the grayscale request according to the loaded fragment configuration information, determine that the grayscale request is the first grayscale request for requesting the grayscale data, or the second grayscale request for requesting the grayscale data and the formal data ;
- a forwarding module configured to forward a portion of the first grayscale request or the second grayscale request requesting the data of the grayscale user to the grayscale data node where the grayscale user corresponding to the grayscale request is located; and the second grayscale request The part requesting official user data is forwarded to the official data node where the official user is located.
- the present application further provides a computing node, including the apparatus for implementing grayscale publishing according to any one of the embodiments of the present application.
- the application further provides a system for implementing grayscale publishing, comprising: a data node cluster in a distributed database, and a cluster of computing nodes in a distributed database; wherein
- the data node cluster in the distributed database includes two or more data nodes configured to carry user data; the data is distributed to the plurality of data nodes through the fragmentation configuration information;
- the computing node cluster in the distributed database includes two or more computing nodes configured to provide external services.
- the computing node includes the device for implementing grayscale publishing according to any one of the embodiments of the present application.
- the technical solution of the present application includes: processing the received application request, determining that the received application request is a grayscale request; parsing the grayscale request according to the loaded fragment configuration information, determining that the grayscale request is requesting grayscale data a first grayscale request, or a second grayscale request requesting grayscale data and official data; performing a portion of the first grayscale request or the second grayscale request requesting grayscale user data according to the grayscale modification configuration information Rewriting; forwarding the portion of the first grayscale request or the second grayscale request requesting the data of the grayscale user to the grayscale data node where the grayscale user corresponding to the grayscale request is located; and the second grayscale request The part requesting official user data is forwarded to the official data node where the official user is located.
- FIG. 1 is a flow chart of a method for implementing grayscale publishing according to the present application
- FIG. 2 is a schematic diagram of an embodiment of implementing data fragmentation and a heterogeneous table of the same name in the present application;
- FIG. 3 is a schematic diagram of an embodiment of user data fragmentation supporting grayscale publishing in the present application.
- FIG. 4 is a schematic diagram of a distributed database data redistribution embodiment of the present application.
- FIG. 5 is a schematic diagram of an embodiment of statistical processing of a grayscale request report according to the present application.
- FIG. 6 is a flowchart of an embodiment of a reconciliation processing of the present application.
- FIG. 7 is a schematic structural diagram of a device for implementing grayscale distribution according to the present application.
- FIG. 8 is a schematic diagram of a composition embodiment of a computing node of the present application.
- FIG. 9 is a schematic diagram of an overall architecture of implementing grayscale publishing based on a distributed database according to the present application.
- FIG. 10 is a schematic structural diagram of a device for implementing grayscale publishing according to the present application.
- FIG. 11 is a schematic diagram of a first embodiment of implementing grayscale distribution according to the present application.
- FIG. 12 is a schematic diagram of a second embodiment of implementing grayscale distribution according to the present application.
- FIG. 13 is a schematic diagram of a third embodiment of implementing grayscale distribution according to the present application.
- FIG. 14 is a schematic diagram of a fourth embodiment of implementing grayscale distribution according to the present application.
- FIG. 1 is a flowchart of a method for implementing grayscale publishing according to the present application. As shown in FIG. 1, the method includes:
- Step 100 The computing node processes the received application request, and determines that the received application request is a grayscale request.
- the method of the present application further includes:
- the formal request is forwarded to the official data node where the official user corresponding to the official request is located.
- the computing node may filter the received application request through the IP whitelist to distinguish whether the received application request is a grayscale request or a formal request for formal data.
- the IP whitelist includes: officially requesting an IP whitelist and a grayscale request IP whitelist; and filtering the received application request by using the IP whitelist includes:
- An application request from a formal application database client in the official request IP whitelist is a formal request
- An application request from a grayscale application database client in the grayscale request IP whitelist is a grayscale request
- This application filters out illegal application requests through the IP whitelist.
- This step also includes:
- Step 101 Parse the grayscale request according to the preset fragment configuration information, determine that the grayscale request is the first grayscale request for requesting the grayscale data, or the second grayscale request for requesting the grayscale data and the official data.
- the fragmentation configuration information may include, but is not limited to, any combination of the following: a list fragment, an ER fragment, a Hash fragment, a Range fragment, and the like.
- the grayscale user is distinguished by a plurality of distribution modes. And official users.
- the application also includes:
- the data segmentation function of the distributed database is used to distinguish the grayscale user data from the official user data at the database layer, that is, to distinguish the grayscale user from the official user.
- the present application uses the same fragment configuration information to ensure that all related data of the same user have the same distribution rule, that is, the same fragment configuration information is used for the data distribution of the same user.
- the same user may involve many tables, but these related tables all use the same fragmentation configuration information.
- the step of parsing the grayscale request according to the preset fragment configuration information in step 101 includes: determining a table to be queried according to the grayscale request; and querying the fragmentation rule of the table according to the fragmentation configuration information. And determining the data node corresponding to the grayscale request according to the fragmentation rule.
- Step 103 Forward a portion of the first grayscale request or the second grayscale request requesting the data of the grayscale user to the grayscale data node where the grayscale user corresponding to the grayscale request is located; and request the second grayscale request Part of the official user data is sent to the official data node where the official user is located.
- the method further includes: sending, by the summary request, the grayscale data node to which the grayscale user is located.
- the execution results of the parts and the execution results of the parts of the official data nodes to which the official users are located are summarized.
- the method for implementing grayscale publishing in the present application further includes:
- Step 102 Rewrite the portion of the first grayscale request or the second grayscale request requesting the grayscale user data according to the preset grayscale modification configuration information.
- the distributed database supports the same-named heterogeneous table, that is, the table with the same name but different data structure.
- the present application also includes:
- the table structure of the grayscale user database is rewritten according to the grayscale modification configuration information, so that the structure of the data table used by the grayscale user and the official user is different.
- the grayscale user's table is different from the formal user's table structure, both still use the same name table.
- the table T is segmented according to the id according to the preset fragment configuration information, and id1 to id10 are distributed on the data node 1, and id11 to id20 are distributed in the data node 2;
- the configuration information refers to which tables, which fields are used to determine whether it needs to be rewritten, and the modified fields are not involved, so the SQL does not need to be rewritten.
- the business uses the same statement to support access to the heterogeneous database without the inconsistency of dimensions.
- the fields in the original table are field A, field B, and field C
- FIG. 2 is a schematic diagram of an embodiment of implementing data fragmentation and a heterogeneous table of the same name in the present application.
- a distributed database When a distributed database is not used, all data of the same table is placed on one data node, and all data rows are stored in the table.
- the table T on a single node stores 9 rows of data.
- data row 1 stores 1 row to 4 rows
- data node 2 stores 5 rows to 8 rows
- data node 3 stores row 9 rows.
- the data can be distributed on different data nodes by framed ranges such as set1 ⁇ 1 ⁇ 4 ⁇ , set2 ⁇ 5 ⁇ 8 ⁇ , set3 ⁇ 9 ⁇ ;
- Another example is if other sharding methods such as hash sharding, relational sharding, etc. are used.
- the table structure on different data nodes can be modified.
- the entire data node still displays the original table structure T (A, B, C) for the calculation node. That is to say, for a distributed database, the table structure is externally displayed by the computing node, and the table T in the formal database and the grayscale database is a whole. Regardless of the grayscale modification, the external table structure is still T ( A, B, C).
- FIG. 3 is a schematic diagram of a user data fragmentation embodiment supporting grayscale publishing according to the present application.
- a selection policy of a grayscale user needs to be configured, that is, a fragmentation policy of a distributed database is preset, for example, it can be set according to a range.
- the user of the 1-100 ID is a grayscale user, which belongs to the Range distribution and divides the data into formal data and grayscale data. Set the same distribution rules for related tables to ensure that all data of the same user will be distributed in the same library.
- the formal user related tables T1, T2, T3 and T4 are all on the formal data nodes, grayscale users.
- the associated table T1, table T2, table T3, and table T4 are all on the grayscale data node.
- the grayscale user's related table needs to perform the corresponding table structure modification, as shown in the grayscale library in Fig. 4, the table name is consistent with the formal data node, but the table structure is different.
- the formal user's query will only involve the official library (data in the dotted line on the left side), and the grayscale user's query will involve the grayscale library and the official library (data in the dotted line on the right).
- Gray-scale users for SQL requests that do not involve gray-listed people after filtering through the static routing gray list, will directly send the SQL statement to the official library, as shown in the basic information table 2 in Figure 4; gray-scale users for the gray-listed population
- the SQL request will be rewritten after the corresponding SQL is sent to the gray node for execution.
- the original select C from T according to the rewrite configuration, will be rewritten as select B+C from T and then sent to the gray Degree node.
- the present application further includes:
- Redistributing data to balance the load including moving part of the hotspot data to the new shard, or migrating from the burdened data node to the less burdened data node.
- User sharding can be re-sharded through the data redistribution function of the distributed database.
- the data redistribution function it is possible to migrate data according to the established fragmentation configuration information without stopping the service.
- FIG. 4 it is assumed that the original data is distributed on four data nodes. However, due to the pressure of the data node 4, the data of the data node 4 can be redistributed, such as migrating a part of the data to the newly added data node. 5 on.
- the method of the present application further includes: modifying a table structure of the corresponding gray scale according to an application characteristic of the grayscale user. For example, in the original database, a certain field M is in units of “yuan”, and the business logic is displayed to the user as “ten thousand yuan”. Then, with grayscale modification, the field in the database will be changed to M/10000.
- the fields in the corresponding official database include the field A, the field B, and the field C
- the fields in the corresponding gray database include the field A, the field B, and the field D
- the computing node receives the general SQL request, firstly, according to the preset fragment configuration information, the data fragment involved is determined. If the user of the official database and the gray database is involved, the SQL rewriting rule is performed according to the SQL. Determine whether it is necessary to rewrite, and finally send the rewritten statement to the corresponding database (DB), that is, the data node to execute.
- DB database
- the method of the present application further includes:
- the data structure of all the original official databases is modified into the data structure of the grayscale database, that is, the data structure in the original official database is upgraded;
- the data structure of the grayscale database is rolled back to the data structure of the official database, that is, the data structure in the grayscale database is rolled back.
- the fields in the corresponding official database include the field A, the field B, and the field C
- the fields in the corresponding gray database include the field A, the field B, and the field D
- the table T2 corresponds to the field in the official database
- the fields in the corresponding grayscale database include field E, field F, and field G.
- the upgrade and fallback scripts are as follows. 3 shows:
- FIG. 5 is a schematic diagram of an embodiment of the grayscale request report statistical processing of the present application, as shown in FIG. 5, including: both grayscale users and formal users.
- the request is generally a statistical report; the formal database and the grayscale database share a set of report generation programs, like a set of stored procedures, the structure of the generated report is the same as the sub-database report; the computing node is responsible for calling the formal database and the gray-scale database report generation program.
- the SQL rewriting rules different parameters are passed to the report generation program to match the table structure of the same-named table in each database; the computing node is responsible for summarizing the reports generated by each database, that is, the sub-database reports.
- both the formal database and the grayscale database have the same set of stored procedures described above.
- different rewrite rules are triggered according to the table related to the report. Incoming different parameters, such as the table T, batch processing the field C in the formal database storage process; processing the result of the field E/10 in the gray database storage process.
- a certain field in the stored procedure statistics table is greater than 100.
- the field C is directly passed in.
- the gray database if you want to implement statistics correctly, you need to pass E/10.
- FIG. 6 is a flow chart of an embodiment of the reconciliation processing of the present application.
- the reconciliation program in the grayscale publishing mainly deals with two problems: only one reconciliation file, and the billing record table is distributed in the official database and gray.
- the structure of the reconciliation records in the formal database and the grayscale database may be different. include:
- the compute node imports the entries in the reconciliation data file into the shard configuration rules of the official user (item 1, entry 2, and entry 3 in Figure 6, and grayscale users (item 4 and entry 5 in Figure 6). To the corresponding temporary table;
- the computing node calls the reconciliation program (such as a stored procedure) in each database, and by loading the SQL rewriting rules, the different parameters are passed to the reconciliation program in the different fragment nodes, and the matching is performed (ie, by rewriting, the request is in the request)
- the statement is adjusted to be the same as the structure of the table in the database) a heterogeneous reconciliation table with the same name in the database;
- the calculation node summarizes the reconciliation results.
- FIG. 7 is a schematic structural diagram of a device for implementing grayscale distribution according to the present application. As shown in FIG. 7, the device includes at least: a filtering module, a parsing module, and a forwarding module;
- a filtering module configured to process the received application request, and determine that the received application request is a grayscale request
- the parsing module is configured to parse the grayscale request according to the preset fragment configuration information, determine that the grayscale request is the first grayscale request for requesting the grayscale data, or request the grayscale data and the second grayscale of the official data. request;
- a forwarding module configured to forward a portion of the first grayscale request or the second grayscale request requesting the data of the grayscale user to the grayscale data node where the grayscale user corresponding to the grayscale request is located; and the second grayscale request The part requesting official user data is forwarded to the official data node where the official user is located.
- the device of the present application further includes:
- the processing module is configured to rewrite the portion of the first grayscale request or the second grayscale request requesting the grayscale user data according to the preset grayscale modification configuration information, and then rewrite the first grayscale request or the second The grayscale request is output to the forwarding module.
- the filtering module is further configured to: if it is determined that the received application request is a formal request for requesting formal data, forward the formal request to the official data node where the official user corresponding to the formal request is located.
- the filtering module is configured to filter the received application request by the IP whitelist to distinguish whether the received application request is a grayscale request for requesting grayscale data or a formal request for formal data. More specifically:
- the IP whitelist is divided into two parts: the official request IP whitelist and the grayscale request IP whitelist; the application request from the official application database client in the official request IP whitelist is a formal request; the gray from the grayscale request IP whitelist Application requests for the application database client are grayscale requests; application requests from non-whitelisted IPs are discarded.
- the parsing module is configured to: determine a table to be queried according to the gradation request; query a sharding rule of the table according to the sharding configuration information, and determine a data node corresponding to the gradation request according to the sharding rule.
- the parsing module is further configured to: according to the pre-set fragment configuration information, use the data fragmentation function of the distributed database to distinguish the gray-scale user data from the official user data in the database layer, that is, Separate grayscale users from formal users. Moreover, the present application ensures that all relevant data of the same user has the same distribution rule by using the same slice configuration information.
- the processing module is further configured to: rewrite the table structure of the grayscale user database according to the grayscale modification configuration information, so that the data table structure used by the grayscale user and the official user is different.
- the processing module is further configured to modify the table structure of the corresponding gray scale according to the application characteristics of the grayscale user.
- the processing module is further configured to: for the second grayscale request involving the grayscale user and the formal user, such as a statistical report, after the data is executed, the grayscale data sent to the grayscale user is sent.
- the portion of the node is aggregated with the execution result data of the portion of the official data node to which the official user is located.
- the processing module is further configured to: redistribute data to balance the load, including: moving part of the hotspot data into the newly added fragment, or migrating from the data node with greater burden to the data with less burden node.
- processing module is further configured to:
- the data structure of all the original official databases is modified into the data structure of the grayscale database, that is, the data structure in the original official database is upgraded; when the user determines that the grayscale application is not applicable, The data structure of the grayscale database is rolled back to the data structure of the official database, that is, the data structure in the grayscale database is rolled back.
- each unit in the apparatus for implementing grayscale distribution may be implemented by a central processing unit (CPU) or a microprocessor (MPU) located in a device that implements grayscale distribution.
- CPU central processing unit
- MPU microprocessor
- Micro Processor Unit Micro Processor Unit
- DSP Digital Signal Processor
- FPGA Field Programmable Gate Array
- FIG. 8 is a schematic diagram of a composition embodiment of a computing node according to the present application.
- the method includes an IP whitelist module, a static routing module, and a data rewriting module.
- the specific implementation of the IP whitelist module is as shown in FIG. A filtering module, the specific implementation of the static routing module is as shown in any one of the parsing modules in FIG. 7, and the specific implementation of the data rewriting module is as shown in any one of the processing modules in FIG.
- the computing node loads the IP whitelist configuration information through the IP whitelist module; the static routing module loads the data fragmentation configuration information; and the SQL data rewriting module loads the SQL rewriting rule configuration information.
- the working principle of the computing node of the present application includes:
- the IP whitelist module of the data computing node performs whitelist filtering on all received external application requests.
- the whitelist is divided into two parts, and the IP whitelist and the grayscale request IP whitelist are formally requested. SQL requests from non-whitelisted IPs will be discarded; requests from official clients (such as IP1 from the figure) will be sent directly to the official database; requests from grayscale clients (from IP2 in the figure) will be sent. Will be sent to the static routing module for processing;
- the static routing module of the computing node routes and parses the SQL request according to the data fragmentation configuration information.
- SQL requests are divided into three types: requesting formal data separately, requesting grayscale data separately, and requesting multiple pieces of user data at the same time.
- the SQL request requesting the official data is sent to the official database; the SQL request requesting the gray data, and the SQL request requesting the multi-part user data at the same time are rewritten by the data rewriting module and sent to the gray database.
- FIG. 9 is a schematic diagram of an overall architecture for implementing grayscale publishing based on a distributed database according to the present application.
- the distributed database front-end docking multiple applications are as shown in FIG. 1, application 1, application 2, application X, and multiple applications respectively. Used to process application requests from official users and application requests from grayscale users.
- the data node cluster in the distributed database includes a plurality of data nodes, such as data node 1, data node 2, and data node M in FIG. 1, for carrying all user data. Data is distributed to multiple data nodes through a fragmentation strategy, and each data node carries part of the user data.
- the computing node cluster in the distributed database includes a plurality of computing nodes, such as the computing node 1, the computing node 2, and the computing node N, for providing services externally, and specifically includes any one of the computing nodes in FIG.
- the compute node adopts a shared-nothing architecture to provide a unified standard structured query language (SQL) interface for applications, and multiple compute nodes perform load balancing.
- the computing node finds the corresponding data storage location according to the data fragmentation information, and can send the corresponding SQL statement to the corresponding data node for execution.
- SQL structured query language
- the present application also proposes a computer readable storage medium having stored thereon a computer program that, when executed by a processor, implements the steps of any of the methods described above for implementing grayscale publishing.
- the device includes a processor and a computer readable storage medium, where the computer readable storage medium stores instructions, where the instructions are When the processor is executed, any of the above methods for realizing grayscale is implemented.
- the computer readable storage medium includes any one or any of the following: a flash memory, a hard disk, a multimedia card, a card type memory (for example, a Secure Digital Memory Card (SD card) or a data register (DX, Data Register) ) Memory, etc., Random Access Memory (RAM), Static Random Access Memory (SRAM), Read Only Memory (ROM), EEPROM (Electrically Erasable Programmable Read Only Memory) EEPROM (Electrically Erasable Programmable Read-Only Memory), Programmable Read-Only Memory (PROM), magnetic memory, magnetic disk, optical disk, and the like.
- a flash memory for example, a Secure Digital Memory Card (SD card) or a data register (DX, Data Register)
- SD card Secure Digital Memory Card
- DX Data Register
- RAM Random Access Memory
- SRAM Static Random Access Memory
- ROM Read Only Memory
- EEPROM Electrically Erasable Programmable Read Only Memory
- EEPROM Electrically Erasable Programmable Read-Only Memory
- PROM
- the processor can be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip.
- CPU central processing unit
- controller a controller
- microcontroller a microcontroller
- microprocessor a microprocessor
- the first embodiment is an embodiment of a multi-level grayscale distribution for a distributed database supporting service.
- FIG. 11 is a schematic diagram of a first embodiment of implementing grayscale publishing according to the present application. As shown in FIG. 11, the distribution in the first embodiment is shown in FIG. There are multiple data nodes in the database, and all data is normally stored on the official library. When the user wants to perform comparison of multiple grayscale modifications at the same time, the data of the grayscale user is distributed to other data nodes such as the grayscale 1 library, the grayscale 2 library, and the grayscale n library. Specifically include:
- the fragmentation information of the table is set according to the user distribution configuration information rule of the grayscale test, and the data is redistributed by using a pre-set fragmentation rule that satisfies the appropriate business: it will be applied to one
- the test user's data of the gray-scale modification is configured to the grayscale 1 library
- the data of the test user applied to the second-type grayscale modification is configured to the grayscale 2 library, ..., which will be applied to the test of the n-type grayscale modification.
- User data is configured to the grayscale library n;
- the static routing module of the computing node parses and routes the original SQL statement, and all the split SQL passes through the data rewriting module, finds the SQL rewriting rule of the corresponding node of the SQL, and rewrites it to the SQL.
- the corresponding grayscale library is executed. Taking the statement select C from T as an example, the statement on the corresponding slice 2 is rewritten as select C-1 from T, and the statement on the corresponding slice 3 is changed to select c-2 from T, ... corresponding to the slice n The statement is rewritten as select cn from T.
- All online single-node transactions (that is, online services involving only one user data) will be returned directly. If the compute node determines that the statement is only sent to node n, selcet c from T where id in set n, the statement will be rewritten to selcet Cn from T where id in set n; batch cross-node transactions (that is, services involving multiple users on multiple shards) are returned by the middle compute nodes, such as compute node judgment statements sent to nodes (n-1) and nodes n,selcet c from T where id in set ⁇ n-1,n ⁇ , the statement will be split into two such as: a statement selcet c-(n-1) from T where id in set n-1 sent to Node (n-1), another statement selcet cn from T where id in set n is sent to node n.
- the upgrade script can be directly generated by the SQL rewriting configuration information corresponding to the modified type of the grayscale node, and the entire database is used by using the upgrade script. Grayscale upgrade.
- the second embodiment is a distributed database supporting global service unified construction.
- FIG. 12 is a schematic diagram of a second embodiment of implementing grayscale distribution according to the present application. As shown in FIG. 12, in the second embodiment, it is assumed that the service design department is responsible for building a unified embodiment.
- the application platform has uniform requirements for data structure to facilitate data report statistics; while the subordinate business construction departments have their own special needs, and the same business is handled differently.
- the distributed database system of the present application can be used to satisfy the need for unified construction of global services. Specifically include:
- the fragmentation information of the table is set according to the user distribution configuration rules of each department, and the data is re-sliced: if the data of the department 1 is configured to the sub-service 1 library, the data of the department 2 is Configured to the sub-service 2 library, ... configures the data of the department n to the sub-service n-bank; each construction department modifies its own table structure according to its own business logic, and executes its own alter statement. On different shards, each department adds a corresponding SQL rewrite configuration rule for the modification operation of the table structure.
- the static routing module of the computing node parses and splits the SQL statement, and all the split SQL requests have corresponding sub-business libraries, and the data rewriting module corresponding to the corresponding business library finds the corresponding The SQL rewrite rules are rewritten and sent to the corresponding sub-business library for execution.
- the third embodiment is to quickly perform the upgrade and rollback of the service using the distributed database.
- FIG. 13 is a schematic diagram of the third embodiment of implementing grayscale distribution according to the present application. As shown in FIG. 13, in the third embodiment, the present embodiment is used. You can apply for the upgrade and rollback methods to quickly upgrade and roll back the service version. This does not take up business time and does not cause service interruption. Specifically include:
- the fourth embodiment is a schematic diagram of the fourth embodiment of implementing grayscale distribution according to the present invention, which uses the SQL automatic rewriting function and the route masking function.
- FIG. 14 in the fourth embodiment, There is no requirement for the number of data nodes in a distributed database. Under normal circumstances, the business is running normally.
- the business logic can be modified by the SQL rewriting function or the SQL masking function provided by the present application to reduce the impact of the bug on the user. Specifically include:
- the logic is more complicated, for example, if the business logic cannot be repaired through SQL modification, the corresponding SQL statement can be masked in the static routing module of the computing node. For example, when the business executes the SQL statement, the error is directly returned, which also limits the error. The scope of the bug impact.
- the method and device for realizing grayscale publishing provided by the present application solve the problem that the application grayscale publishing needs to continuously modify the data structure and the business logic without modifying the business code, and all the modification operations are merged into the database for execution. And through the function of distributed data redistribution, the business department can construct the grayscale business data more quickly, which saves the system resource investment and improves the service grayscale iteration efficiency.
Abstract
Description
Claims (28)
- 一种实现灰度发布的方法,包括:对接收到的应用请求进行处理,确定出接收到的应用请求为灰度请求;根据预先设置的分片配置信息对灰度请求进行解析,确定灰度请求是请求灰度数据的第一灰度请求,或者是请求灰度数据和正式数据的第二灰度请求;将第一灰度请求或第二灰度请求中请求灰度用户的数据的部分转发至灰度请求对应的灰度用户所在的灰度数据节点执行;将第二灰度请求中请求正式用户数据的部分转发至正式用户所在的正式数据节点执行。
- 根据权利要求1所述的方法,所述方法还包括:根据预先设置的所述分片配置信息,利用分布式数据库的数据分片功能,在数据库层区分所述灰度用户的数据和所述正式用户的数据;其中,同一用户使用相同的分片配置信息。
- 根据权利要求1所述的方法,所述根据预先设置的分片配置信息对灰度请求进行解析之后,所述执行之前,还包括:根据预先设置的灰度修改配置信息对所述第一灰度请求或所述第二灰度请求中请求灰度用户数据的部分进行改写。
- 根据权利要求3所述的方法,所述方法还包括:根据所述灰度修改配置信息对灰度用户数据库的表结构进行改写,使得所述灰度用户和所述正式用户使用的数据表结构不一样,但使用相同名称。
- 根据权利要求1~4任一项所述的方法,如果所述确定出接收到的应用请求是请求正式数据的正式请求,还包括:将正式请求转发至正式请求对应的正式用户所在的正式数据节点执 行。
- 根据权利要求1所述的方法,其中,通过预先设置的IP白名单对所述接收到的应用请求进行过滤,来区分所述接收到的应用请求是请求灰度数据的灰度请求,还是请求正式数据的正式请求。
- 根据权利要求6所述的方法,其中,所述IP白名单包括:正式请求IP白名单和灰度请求IP白名单;所述通过IP白名单对接收到的应用请求进行过滤包括:确定来自正式请求IP白名单中的正式应用数据库客户端的应用请求为正式请求;确定来自灰度请求IP白名单中的灰度应用数据库客户端的应用请求为所述灰度请求;丢弃来自非白名单IP的应用请求。
- 根据权利要求1~4任一项所述的方法,其中,所述根据加载的分片配置信息对灰度请求进行解析包括:根据所述灰度请求确定需要查询的表;根据所述分片配置信息查询该表的分片规则,并按照分片规则确定该灰度请求对应的数据节点。
- 根据权利要求1~4任一项所述的方法,所述第二灰度请求的数据执行完后,还包括:将发往所述灰度用户所在的灰度数据节点的部分的执行结果,以及发往所述正式用户所在的正式数据节点的部分的执行结果进行汇总。
- 根据权利要求1~4任一项所述的方法,所述方法还包括:对数据进行重新分布,将部分热点数据移动到新增分片中,或者将部分热点数据从负担大的数据节点迁移到负担小的数据节点。
- 根据权利要求1~4任一项所述的方法,所述方法还包括:根据所述灰度用户的应用特性,修改对应灰度表的表结构。
- 根据权利要求1~4任一项所述的方法,所述方法还包括:当判断出灰度应用正式发布时,将所有正式数据库的数据结构修改为所述灰度应用对应的灰度数据库的数据结构;当判断出灰度应用不适用时,将所述灰度应用对应的灰度数据库的数据结构回退为正式数据库的数据结构。
- 一种计算机可读存储介质,其上存储有计算机程序,所述计算机程序被处理器执行时实现上述权利要求1~权利要求12任意一种实现灰度发布的方法的步骤。
- 一种实现灰度发布的设备,包括处理器和计算机可读存储介质,所述计算机可读存储介质中存储有指令,当所述指令被所述处理器执行时,实现权利要求1~权利要求12任意一种实现灰度发布的方法。
- 一种实现灰度发布的装置,包括:过滤模块、解析模块,以及转发模块;其中,过滤模块,配置为对接收到的应用请求进行处理,确定出接收到的应用请求为灰度请求;解析模块,配置为根据加载的分片配置信息对灰度请求进行解析,确定灰度请求是请求灰度数据的第一灰度请求,或者是请求灰度数据和正式数据的第二灰度请求;转发模块,配置为将第一灰度请求或第二灰度请求中请求灰度用户的数据的部分转发至灰度请求对应的灰度用户所在的灰度数据节点执行;将第二灰度请求中请求正式用户数据的部分转发至正式用户所在的正式数据节点执行。
- 根据权利要求15所述的装置,所述装置还包括:处理模块,配置为根据所述灰度修改配置信息对所述第一灰度请求或所述第二灰度请求中请求灰度用户数据的部分进行改写,再将改写后 的第一灰度请求或第二灰度请求输出给所述转发模块。
- 根据权利要求15所述的装置,所述过滤模块还配置为:如果确定出所述接收到的应用请求是请求正式数据的正式请求,将正式请求转发至正式请求对应的正式用户所在的正式数据节点执行。
- 根据权利要求15、16或17所述的装置,其中,所述过滤模块用于通过IP白名单对所述接收到的应用请求进行过滤,区分所述接收到的应用请求是请求灰度数据的灰度请求,还是请求正式数据的正式请求。
- 根据权利要求15、16或17所述的装置,其中,所述IP白名单包括:正式请求IP白名单和灰度请求IP白名单;所述区分所述接收到的应用请求是请求灰度数据的灰度请求,还是请求正式数据的正式请求包括:确定来自正式请求IP白名单中的正式应用数据库客户端的应用请求为正式请求;确定来自灰度请求IP白名单中的灰度应用数据库客户端的应用请求为灰度请求;丢弃来自非白名单IP的应用请求。
- 根据权利要求15所述的装置,其中,所述解析模块配置为:根据所述灰度请求确定需要查询的表;根据所述分片配置信息查询该表的分片规则,并按照分片规则确定该灰度请求对应的数据节点。
- 根据权利要求20所述的装置,所述解析模块还配置为:根据预先设置的所述分片配置信息,利用分布式数据库的数据分片功能,在数据库层区分灰度用户的数据和正式用户的数据;其中,同一用户使用相同的分片配置信息。
- 根据权利要求15所述的装置,所述处理模块还配置为:根据所述灰度修改配置信息对灰度用户数据库的表结构进行改写,使得所述灰度用户和所述正式用户使用的数据表结构不一样。
- 根据权利要求15所述的装置,所述处理模块还配置为:根据所 述灰度用户的应用特性,修改对应灰度表的表结构。
- 根据权利要求15所述的装置,所述处理模块还配置为:所述第二灰度请求在数据执行完后,将发往所述灰度用户所在的灰度数据节点的部分的执行结果,以及发往正式用户所在的正式数据节点的部分的执行结果进行汇总。
- 根据权利要求15所述的装置,所述处理模块还配置为:对数据进行重新分布,将部分热点数据移动到新增分片中,或者将部分热点数据从负担大的数据节点迁移到负担小的数据节点。
- 根据权利要求15所述的装置,所述处理模块还配置为:当判断出灰度应用正式发布时,将所有正式数据库的数据结构修改为所述灰度应用对应的灰度数据库的数据结构;当判断出灰度应用不适用时,将所述灰度应用对应的灰度数据库的数据结构回退为正式数据库的数据结构。
- 一种计算节点,包括权利要求15~26任一项所述的实现灰度发布的装置。
- 一种实现灰度发布的系统,包括:分布式数据库中的数据节点集群、分布式数据库中的计算节点集群;其中,分布式数据库中的数据节点集群包括两个或两个以上数据节点,配置为承载用户数据;数据通过分片配置信息分布到多个数据节点;分布式数据库中的计算节点集群包括两个或两个以上计算节点,配置为对外提供服务,所述计算节点包括权利要求15~26任一项所述的实现灰度发布的装置。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2020536074A JP7083901B2 (ja) | 2017-12-29 | 2018-07-23 | ダークローンチ実現方法、装置、計算ノード及びシステム |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711499192.5 | 2017-12-29 | ||
CN201711499192.5A CN108319656A (zh) | 2017-12-29 | 2017-12-29 | 实现灰度发布的方法、装置及计算节点和系统 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2019128205A1 true WO2019128205A1 (zh) | 2019-07-04 |
Family
ID=62892765
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2018/096686 WO2019128205A1 (zh) | 2017-12-29 | 2018-07-23 | 实现灰度发布的方法、装置及计算节点和系统 |
Country Status (3)
Country | Link |
---|---|
JP (1) | JP7083901B2 (zh) |
CN (1) | CN108319656A (zh) |
WO (1) | WO2019128205A1 (zh) |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111352654A (zh) * | 2020-03-10 | 2020-06-30 | 中国工商银行股份有限公司 | 用于实现灰度发布的系统和方法 |
CN111737128A (zh) * | 2020-06-19 | 2020-10-02 | 北京百度网讯科技有限公司 | 线上测试方法及灰度分流设备、存储介质 |
CN111897571A (zh) * | 2020-08-04 | 2020-11-06 | 上海非码网络科技有限公司 | 基于Spring Cloud的灰度发布方法及其系统 |
CN113138774A (zh) * | 2021-05-11 | 2021-07-20 | 中国工商银行股份有限公司 | 灰度发布方法、装置、电子设备、介质和程序产品 |
CN113296911A (zh) * | 2021-05-24 | 2021-08-24 | 北京京东振世信息技术有限公司 | 集群调用方法、集群调用装置、电子设备及可读存储介质 |
CN113535595A (zh) * | 2021-09-17 | 2021-10-22 | 国网江苏省电力有限公司营销服务中心 | 基于微服务架构的信息系统灰度发布用户选择方法和系统 |
CN113626068A (zh) * | 2020-05-09 | 2021-11-09 | 北京沃东天骏信息技术有限公司 | 一种页面管理方法和装置 |
CN114697304A (zh) * | 2020-12-28 | 2022-07-01 | 京东科技控股股份有限公司 | 一种灰度发布方法、系统、装置、设备及存储介质 |
CN114944951A (zh) * | 2022-05-18 | 2022-08-26 | 北京天融信网络安全技术有限公司 | 一种请求的处理方法及装置、拟态设备、存储介质 |
CN117315107A (zh) * | 2023-09-21 | 2023-12-29 | 支付宝(杭州)信息技术有限公司 | 数字形象发布方法及装置 |
Families Citing this family (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109189494B (zh) * | 2018-07-27 | 2022-01-21 | 创新先进技术有限公司 | 配置灰度发布方法、装置、设备及计算机可读存储介质 |
CN110225094B (zh) * | 2019-05-17 | 2022-02-11 | 平安科技(深圳)有限公司 | 负载均衡应用虚拟ip切换方法、装置、计算机设备及存储介质 |
CN110489133A (zh) * | 2019-08-23 | 2019-11-22 | 亿企赢网络科技有限公司 | 一种灰度发布方法、系统及电子设备和存储介质 |
CN110532008A (zh) * | 2019-08-29 | 2019-12-03 | 深圳前海环融联易信息科技服务有限公司 | 一种产品灰度发布方法、装置、计算机设备及存储介质 |
CN111158715B (zh) * | 2019-12-19 | 2024-02-06 | 中国平安财产保险股份有限公司 | 灰度发布控制方法及系统 |
CN111580859B (zh) * | 2020-05-12 | 2023-04-07 | 北京思特奇信息技术股份有限公司 | 一种灰度割接方法、系统及电子设备 |
CN114780512B (zh) * | 2022-03-22 | 2023-05-12 | 荣耀终端有限公司 | 一种灰度发布方法、系统及服务器 |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105791341A (zh) * | 2014-12-22 | 2016-07-20 | 华为软件技术有限公司 | 一种应用发布的处理方法、装置及系统 |
CN106909563A (zh) * | 2015-12-23 | 2017-06-30 | 上海热璞网络科技有限公司 | 一种分布式系统 |
US20170193061A1 (en) * | 2015-12-30 | 2017-07-06 | Alexander Visbal | Systems and methods for search template generation |
Family Cites Families (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CA2438997A1 (en) | 2003-08-28 | 2005-02-28 | Ibm Canada Limited - Ibm Canada Limitee | System and method for carrying out legacy application transitions |
JP5119027B2 (ja) | 2008-03-31 | 2013-01-16 | 株式会社野村総合研究所 | 情報処理装置およびプログラム解析方法 |
JP4931881B2 (ja) | 2008-08-13 | 2012-05-16 | 日本電信電話株式会社 | ホワイトリストを利用したサーバ割り当てシステムおよびその方法 |
US9113130B2 (en) * | 2012-02-06 | 2015-08-18 | Legend3D, Inc. | Multi-stage production pipeline system |
JP2017033136A (ja) | 2015-07-30 | 2017-02-09 | 富士通株式会社 | 情報制御プログラム、情報制御装置及び情報制御方法 |
CN106844399B (zh) * | 2015-12-07 | 2022-08-09 | 中兴通讯股份有限公司 | 分布式数据库系统及其自适应方法 |
US10127272B2 (en) | 2016-03-22 | 2018-11-13 | Oracle International Corporation | Modifying a database query |
CN105978976A (zh) * | 2016-05-13 | 2016-09-28 | 中国建设银行股份有限公司 | 信息处理系统及方法 |
-
2017
- 2017-12-29 CN CN201711499192.5A patent/CN108319656A/zh active Pending
-
2018
- 2018-07-23 JP JP2020536074A patent/JP7083901B2/ja active Active
- 2018-07-23 WO PCT/CN2018/096686 patent/WO2019128205A1/zh active Application Filing
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105791341A (zh) * | 2014-12-22 | 2016-07-20 | 华为软件技术有限公司 | 一种应用发布的处理方法、装置及系统 |
CN106909563A (zh) * | 2015-12-23 | 2017-06-30 | 上海热璞网络科技有限公司 | 一种分布式系统 |
US20170193061A1 (en) * | 2015-12-30 | 2017-07-06 | Alexander Visbal | Systems and methods for search template generation |
Cited By (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111352654A (zh) * | 2020-03-10 | 2020-06-30 | 中国工商银行股份有限公司 | 用于实现灰度发布的系统和方法 |
CN111352654B (zh) * | 2020-03-10 | 2023-08-15 | 中国工商银行股份有限公司 | 用于实现灰度发布的系统和方法 |
CN113626068A (zh) * | 2020-05-09 | 2021-11-09 | 北京沃东天骏信息技术有限公司 | 一种页面管理方法和装置 |
CN111737128A (zh) * | 2020-06-19 | 2020-10-02 | 北京百度网讯科技有限公司 | 线上测试方法及灰度分流设备、存储介质 |
CN111897571A (zh) * | 2020-08-04 | 2020-11-06 | 上海非码网络科技有限公司 | 基于Spring Cloud的灰度发布方法及其系统 |
CN114697304B (zh) * | 2020-12-28 | 2023-12-01 | 京东科技控股股份有限公司 | 一种灰度发布方法、系统、装置、设备及存储介质 |
CN114697304A (zh) * | 2020-12-28 | 2022-07-01 | 京东科技控股股份有限公司 | 一种灰度发布方法、系统、装置、设备及存储介质 |
WO2022142536A1 (zh) * | 2020-12-28 | 2022-07-07 | 京东科技控股股份有限公司 | 灰度发布方法、系统、装置、设备及存储介质 |
CN113138774A (zh) * | 2021-05-11 | 2021-07-20 | 中国工商银行股份有限公司 | 灰度发布方法、装置、电子设备、介质和程序产品 |
CN113138774B (zh) * | 2021-05-11 | 2024-02-09 | 中国工商银行股份有限公司 | 灰度发布方法、装置、电子设备和介质 |
CN113296911A (zh) * | 2021-05-24 | 2021-08-24 | 北京京东振世信息技术有限公司 | 集群调用方法、集群调用装置、电子设备及可读存储介质 |
CN113296911B (zh) * | 2021-05-24 | 2023-09-22 | 北京京东振世信息技术有限公司 | 集群调用方法、集群调用装置、电子设备及可读存储介质 |
CN113535595A (zh) * | 2021-09-17 | 2021-10-22 | 国网江苏省电力有限公司营销服务中心 | 基于微服务架构的信息系统灰度发布用户选择方法和系统 |
CN114944951A (zh) * | 2022-05-18 | 2022-08-26 | 北京天融信网络安全技术有限公司 | 一种请求的处理方法及装置、拟态设备、存储介质 |
CN114944951B (zh) * | 2022-05-18 | 2024-02-06 | 北京天融信网络安全技术有限公司 | 一种请求的处理方法及装置、拟态设备、存储介质 |
CN117315107A (zh) * | 2023-09-21 | 2023-12-29 | 支付宝(杭州)信息技术有限公司 | 数字形象发布方法及装置 |
CN117315107B (zh) * | 2023-09-21 | 2024-03-12 | 支付宝(杭州)信息技术有限公司 | 数字形象发布方法及装置 |
Also Published As
Publication number | Publication date |
---|---|
CN108319656A (zh) | 2018-07-24 |
JP7083901B2 (ja) | 2022-06-13 |
JP2021513694A (ja) | 2021-05-27 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2019128205A1 (zh) | 实现灰度发布的方法、装置及计算节点和系统 | |
JP6815456B2 (ja) | 複数ソースからのデータの処理 | |
US20210182241A1 (en) | Utilizing metadata to prune a data set | |
US11341139B2 (en) | Incremental and collocated redistribution for expansion of online shared nothing database | |
US9589041B2 (en) | Client and server integration for replicating data | |
US8738568B2 (en) | User-defined parallelization in transactional replication of in-memory database | |
US10007698B2 (en) | Table parameterized functions in database | |
US9529881B2 (en) | Difference determination in a database environment | |
US10503727B2 (en) | System and method for mapping database changes | |
US8949222B2 (en) | Changing the compression level of query plans | |
US9256642B2 (en) | Techniques for recommending parallel execution of SQL statements | |
US7885953B2 (en) | Off-loading star join operations to a storage server | |
US10719506B2 (en) | Natural language query generation | |
WO2015167466A1 (en) | Query plan post optimization analysis and reoptimization | |
US9524305B2 (en) | Boolean term conversion for null-tolerant disjunctive predicates | |
US11055289B2 (en) | Framework to transfer rows in storage format, standard format, or combination thereof | |
WO2023045295A1 (zh) | 数据倾斜处理方法、设备、存储介质及程序产品 | |
US20130339396A1 (en) | Asynchronously flattening graphs in relational stores | |
US11232000B1 (en) | Moving database partitions from replica nodes | |
CN105653334B (zh) | 一种基于saas模式的mis系统快速开发框架 | |
EP3462341B1 (en) | Local identifiers for database objects | |
Li et al. | Apache shardingsphere: A holistic and pluggable platform for data sharding | |
CN113268483A (zh) | 请求处理方法和装置、电子设备和存储介质 | |
US20150033232A1 (en) | Automatic parallelism tuning for apply processes | |
US20230072930A1 (en) | Database query splitting |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
ENP | Entry into the national phase |
Ref document number: 2020536074 Country of ref document: JP Kind code of ref document: A |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 18893416 Country of ref document: EP Kind code of ref document: A1 |
|
32PN | Ep: public notification in the ep bulletin as address of the adressee cannot be established |
Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 19.11.2020) |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 18893416 Country of ref document: EP Kind code of ref document: A1 |