Patent Application
20180025066
It is important for database systems that service client requests to be highly available and responsive to such requests. For example, database clients in a high-availability database environment (e.g., a web-based or cloud database environment supporting many clients) may expect that the database system be continuously available with little, if any, downtime. Furthermore, database clients may expect that response times for database requests be very short, such as less than a second or a small number of seconds.
In such high-availability database environments, any downtime or interruption in service can be a problem. For example, in some database systems, the database is taken offline when changes to the database need to be made. Taking a database offline can result in dropped or failed connections, timeouts, delays, or other problems connecting to the database. In some database systems, database requests are queued while changes to the database need to be made. Queueing database requests can result in delayed responses, failed connections, timeouts, or other degradations in service.
Therefore, there exists ample opportunity for improvement in technologies related to performing updates to database systems.
This Summary is provided to introduce a selection of concepts in a simplified form that are further described below in the Detailed Description. This Summary is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used to limit the scope of the claimed subject matter.
Various technologies are described herein for performing database updates in a database environment where the database remains accessible during the updates. For example, a number of database clients (e.g., computer servers, client devices, etc.) may utilize a database environment for running queries or performing transactions. When an update to the database environment needs to be performed (e.g., a change to the database schema), instead of taking the database offline, the database can remain accessible to the database clients throughout the update process. The update process can include switching database connections between a first copy of the database and a second copy of the database that are being replicated. Technologies are also described for performing updates to application code used to interact with the database environment (e.g., alone or in combination with database system updates).
The following description is directed to technologies for performing database updates (database system changes) in a database environment where the database remains accessible during the updates, which is also called zero downtime database updates. For example, a number of database clients (e.g., computer servers, client devices, etc.) may utilize a database environment for running queries or performing transactions. When an update to the database environment needs to be performed (e.g., a change to the database schema), instead of taking the database offline, the database can remain accessible to the database clients throughout the update process. Furthermore, the database environment can remain accessible without queueing the database requests (e.g., queries or transactions) or otherwise interrupting the database requests during the update process. Technologies are also described for performing updates to application code used to interact with the database environment (e.g., alone or in combination with database system updates).
For example, database updates can be performed in a database environment where a database is being replicated from a first copy of the database (e.g., a current master database) to at least one second copy of the database (e.g., a current slave database). Database connections from database clients can be directed to the first copy of the database. The database system changes can be performed on the at least one second copy of the database. Once the database system changes have been performed, the database connections can be switched from the first copy of the database to the second copy of the database and replication can be started from the second copy to the first copy (e.g., the replication can be reversed).
Database system changes refer to changes to the database itself (e.g., database schema changes), changes to the configuration of the database, or hardware or software changes. For example, database schema changes include adding columns to a table. Database configuration changes include changes to parameters of the database (e.g., buffer settings, log file settings, etc.) or other changes such as rebooting the database. Hardware changes include changes to computing device hardware on which the database is running (e.g., adding ram or storage, replacing a hardware component, replacing a failed hard drive, etc.). Software changes include changes to software running on the computing device (e.g., changes to the operating system or virtual machine) other that the database software.
Application code used to interact with the database environment can also be updated, separately or in combination with updates to the database environment. For example, a current code version can be currently deployed on server and client devices. A new code version can then be deployed and activated once the updates to the database environment are complete.
In some implementations, returned database data (e.g., rows of data returned by queries) is stored in a memory cache (also called an in-memory cache process). For example, the memory cache can be implemented using the Memcached system (a memory cache system described at https://memcached.org/) or by other technology that caches database results in memory (e.g., using a distributed system that stores the data in memory of multiple devices). The returned database data can be serialized into a key-value pair and stored in the memory cache. In some implementations, the keys are modified to include an indication of the code version used to create the data associated with the keys. For example, database data that is retrieved using server applications running a current code version can store key-value data in the memory cache where the keys have an indication of the current code version stored in the key (e.g., a build number, commit identifier, or other indication can be added to the beginning or end of the key). When the server applications are updated to a new code version, new keys can include an indication of the new code version. For example, database data that is retrieved using server applications running the new code version can store key-value data in the memory cache where the keys have an indication of the new code version stored in the key. In this way, the key-value data stored in the memory cache by the current code version can be distinguished from the key-value data stored in the memory cache by the new code version.
Being able to distinguish the data in the memory cache between the code versions can be important when the code version of the server applications is being updated. For example, a new code version of the server applications may not operate correctly with data that was stored in the memory cache by an older code version. Therefore, server applications running a given code version can be configured to operate only with keys associated with the given code version in the memory cache.
In a high-availability database environment (e.g., a web-based or cloud database environment supporting may clients), taking the database offline for configuration changes can be problematic. In some solutions, configuration changes are handled by queueing database operations (e.g., write operations) during the changes. While queueing database operations can maintain availability of the database, there are a number of problems. For example, queueing database operations may only allow the database to be taken offline for a short time period (e.g., a number of seconds), which may not be sufficient time to make significant changes to the database system. In addition, database software and/or application software may need to be re-written in order to support queueing of database operations (e.g., to queue operations in a memory queue until the database becomes available). Finally, queueing database operations may result in a delayed response to database clients and/or database errors.
In the technologies described herein, updates to database environments and/or updates to application code used to interact with the database environments can be performed while the database remains accessible and without queueing database operations during the update procedure. For example, database operations can continue, including write operations that add, update, and/or delete database data, while the updates are performed. By maintaining access to the database during the update procedure, computing resources can be saved (e.g., memory or storage that would otherwise be used by queueing database operations). In addition, savings in terms of responsiveness can also be realized (e.g., operations do not have to be delayed in the queue and can instead be completed more quickly). Finally, various combinations of operations used to perform the update procedure can minimize the chance of a problem occurring within the database environment during the update procedure (e.g., minimize the chance of any database writes being lost or database connections failing).
The technologies described herein can be applied to perform zero downtime database updates in a database environment such as a MySQL® database (MySQL is a database system provided by Oracle® Corporation), an SAP HANA® database (SAP HANA is a database platform provided by SAP®), or another type of database.
In the example database environment, the first copy of the database 140 is being replicated to the second copy of the database 142, as indicated at 152. The first copy of the database 140 may also be replicated to the third copy of the database 144. In some implementations, the first copy of the database 140 is in a read-write mode while the second copy of the database 142 and the third copy of the database 144 are in a read-only mode.
The database environment services database requests from client devices (e.g., client devices 110 and 112) which are connected via a network 115 (e.g., comprising the Internet) and routed through application servers (e.g., application servers 120 and 122). In some implementations, the client devices are running web browser database applications that connected to web server applications running on the application servers. In some implementations, the client devices can connect directly to the database environment.
In the database environment, database connections from the application servers and/or client devices are controlled by functionality, depicted at 130, for directing database connections to a current database. For example, the functionality 130 can be a load balancer, router, networking device, Internet protocol (IP) address service, a virtual component, or another type of computing device or service configured to direct network connections to particular copies of the database. The functionality 130 supports switching between copies of the database when needed.
In some implementations, the functionality 130 controls a virtual IP address to which the application servers and/or client devices connect. The functionality 130 then directs the database connections to one of the database copies. As depicted in
The environment supports updating of the database environment, including database system changes, while the database remains accessible, as depicted at 125. In order to perform the update process, a number of operations are performed. The operations are typically performed in a specific order, but the order of some of the operations can vary while still achieving the result of a zero downtime database update. At the beginning of the update process, database connections are directed to the first copy of the database 140, as depicted at 151, and the first copy of the database 140 is being replicated to the second copy of the database 142 (and in some implementations to the third copy of the database 144 as well), as depicted at 152. Next, the replication from the first copy of the database 140 to the second copy of the database 142 is stopped (and in some implementations to the third copy of the database 144 as well), as depicted at 153. After the replication stops, database system changes are performed to the second copy of the database 142, as depicted at 153. The database system changes can comprise changes to the database schema, changes to the database configuration, hardware and/or software changes to the computing device on which the database is running, and/or other changes related to the second copy of the database 142. After the database system changes are complete, replication from the first copy of the database 140 to the second copy of the database 142 is resumed. For example, during the time period in which the database system changes are performed (e.g., a number of seconds, minutes, or hours) database writes are still being directed to the first copy of the database 140. In order to prepare for switching the database connections and the replication, the updated database data needs to be replicated so that the first and second copies are again in sync.
The update process continues in
As depicted at 125, the entire updated process can be performed while the database remains accessible to the application servers and client devices. For example, the application servers and client devices can continue to perform database operations, via database connections managed by the functionality 130, including operations that read and/or write database data.
The environment also supports updating of code versions of software run on the application servers (e.g., application servers 120 and 122) and/or client devices (e.g., client devices 110 and 112), collectively called server applications. For example, the application servers and/or client devices can be running a first code version during the update process. The first code version (also called a current code version) is configured to operate with the database before the database configuration changes have been applied as well as after the database configuration changes have been applied. For example, the first code version can operate with the database even after database schema changes have been made (e.g., rows inserted using the first code version can use default field values for any new columns added in the schema change). As part of the update process a second code version (also called a new code version) can be introduced. In some implementations, the new code version is deployed (but remains inactive) to the application servers and/or client devices at or near the beginning of the update process (e.g., before the replication between the first copy of the database 140 and the second copy of the database 142 is stopped). However, the new code version can be deployed at other times as well. The new code version can be activated once the database connections are directed to the second copy of the database 142. In some implementations, the new code version is activated once the operations depicted at 155 are complete.
For example, the new code version can be deployed in combination with the database system changes in order to take advantage of database schema changes or other changes to the database system. For example, the new code version can utilize new database columns added by a schema change or remove support for database columns that have been deleted. The new code version can also be deployed independent of any changes to the database system.
Updating the code version can involve making changes to memory cache keys. For example, a memory cache system can be implemented on one or more of the application servers (e.g., a distributed memory cache system running on application servers 120 and 122). Database data that is retrieved from the database by server applications running a first code version can be stored in the memory cache using keys that include an indication of the first code version. When the server applications are updated to a second code version, database data that is retrieved from the database by server applications running the second code version can be stored in the memory cache using keys that include an indication of the second code version. Therefore, the database data stored in the memory cache can be separated based on the code version. In addition, the code version update can be performed while the database environment and memory cache remain accessible (e.g., the memory cache does not have to be restarted or cleared in order to update the code version).
In the example database environment, the master copy of the database 240 is being replicated to the slave copy of the database 242, as indicated at 252. The master copy of the database 240 may also be replicated to a second slave copy of the database 244. In some implementations, the master copy of the database 240 is in a read-write mode while the slave copies, 242 and 244, are in a read-only mode.
The database environment services database requests from client devices (e.g., client devices 210 and 212) which are connected via a network 215 (e.g., comprising the Internet) and routed through web application servers (e.g., web application servers 220 and 222). The client devices access the web application servers via web browsers running on the client devices. In some implementations, the client devices can connect to the database environment using software other than web browsers (e.g., via client database applications).
In the database environment, database connections from the web application servers connect to a virtual IP address. The virtual IP address is then directed, by a component 230, to one of the copies of the database. For example, the component 230 can be a load balancer, router, networking device, virtual component, or another type of computing device or service configured to direct network connections to particular copies of the database. The component 230 supports switching between copies of the database when needed. As depicted in
The environment supports updating of the database environment, including database system changes, while the database remains accessible, as depicted at 225. In order to perform the update process, a number of operations are performed. The operations are typically performed in a specific order, but the order of some of the operations can vary while still achieving the result of a zero downtime database update. At the beginning of the update process, database connections are directed to the master copy of the database 240, as depicted at 251, and the master copy of the database 240 is being replicated to the slave copy of the database 242 (and in some implementations to the second slave copy of the database 244 as well), as depicted at 252. Next, a new code version of the server applications is deployed. The new code version remains inactive. For example, the new code version can be deployed to the web application servers and retrieved by the client devices when needed (e.g., when the client devices first need to run the new code version).
Next, the replication from the master copy of the database 240 to the slave copy of the database 242 is stopped (and in some implementations to the second slave copy of the database 244 as well), as depicted at 253. After the replication stops, database system changes are performed to the slave copy of the database 242, as depicted at 253. The database system changes can comprise changes to the database schema, changes to the database configuration, hardware and/or software changes to the computing device on which the database is running, and/or other changes related to the slave copy of the database 242. After the database system changes are complete, replication from the master copy of the database 240 to the slave copy of the database 242 is resumed, as depicted at 254. For example, during the time period in which the database system changes are performed (e.g., a number of seconds, minutes, or hours) database writes are still being directed to the master copy of the database 240. In order to prepare for switching the database connections and the replication, the updated database data needs to be replicated so that the master and slave copies are again in sync.
The update process continues in
In some implementations, switching to the new master copy of the database 242 comprises a number of operations. First, the new master copy of the database 242 is set to a read-write mode. Next, the virtual IP address is redirected, via the component 230, to direct database connections to the new master copy of the database 242. Next, replication from the new slave copy of the database 240 to the new master copy of the database 242, which continued until this point, is stopped. Next, the new slave copy of the database 240 is set to a read-only mode. Next, replication is started from the new master copy of the database 242 to the new slave copy of the database 240. Finally, the new code version previously deployed to the web application servers is activated and used by the web application servers and client devices.
As depicted at 225, the entire updated process can be performed while the database remains accessible to the web application servers and client devices. For example, the web application servers and client devices can continue to perform database operations, via database connections managed by the component 230, including operations that read and/or write database data.
The environment supports updating of code versions of software run on the web application servers (e.g., web application servers 220 and 222) and/or client devices (e.g., client devices 210 and 212), collectively called server applications. For example, the web application servers and/or client devices can be running a first code version during the update process. The first code version can utilize memory caches (e.g., memory caches 221 and 223) that store database data with keys associated with the first code version. The first code version (also called a current code version) is configured to operate with the database before the database configuration changes have been applied as well as after the database configuration changes have been applied. For example, the first code version can operate with the database even after database schema changes have been made (e.g., rows inserted using the first code version can use default field values for any new columns added in the schema change). As part of the update process a second code version (also called a new code version) can be introduced. In some implementations, the new code version is deployed (but remains inactive) to the web application servers and/or client devices at or near the beginning of the update process. However, the new code version can be deployed at other times as well. The new code version can utilize the memory caches (e.g., memory caches 221 and 223) that store database data with keys associated with the new code version. The new code version can be activated once the database connections are directed to the new master copy of the database 242.
At 320, replication from the first copy of the database to the second copy of the database is stopped.
At 330, database system changes are performed to the second copy of the database. The database system changes can comprise database schema changes to the second copy of the database. The database system changes can also comprise changes to the database configuration, hardware and/or software changes to the device (e.g., computer server) on which the second copy of the database is running, and/or other changes associated with the second copy of the database.
At 340, replication is resumed from the first copy of the database to the second copy of the database.
At 350, database connections are switched so that they are directed to the second copy of the database instead of the first copy of the database. In some implementations, database connections that perform write operations are switched, while database connections that do not perform write operations may be directed to the first or second copy of the database.
At 360, replication from the first copy of the database to the second copy of the database is stopped and replication is started from the second copy of the database to the first copy of the database.
At 420, a new code version of the server applications is deployed. The new code version remains in an inactive state until it is activated.
At 430, replication from the master copy of the database to the slave copy of the database is stopped.
At 440, database system changes are performed to the slave copy of the database. The database system changes can comprise database schema changes to the slave copy of the database. The database system changes can also comprise changes to the database configuration, hardware and/or software changes to the device (e.g., computer server) on which the slave copy of the database is running, and/or other changes associated with the slave copy of the database.
At 450, replication is resumed from the first copy of the database to the second copy of the database.
At 460, once the master and slave copies are in sync after the resumed replication, the copies of the database are switched so that the slave copy becomes the new master copy of the database and the master copy becomes the new slave copy of the database. Switching the copies of the database includes directing database connections (e.g., with write operations) to the new master copy of the database.
In some implementations, switching the copies of the database at 460 involves a number of operations. First, the new master copy of the database is set to a read-write mode. Next, a virtual IP address is switched to point to the new master copy of the database. Next, replication is stopped from the new slave copy of the database to the new master copy of the database. Next, the new slave copy of the database is set to a read-only mode. Next, replication is started from the new master copy of the database to the new slave copy of the database. Finally, the new code version of the server applications is activated.
In an example implementation, zero downtime database updates are performed according to specific conditions and using a specific sequence of operations. The conditions and operations described in this section can be used separately or in combination with other techniques described herein.
In the example implementation, a number of preconditions are established. According to a first precondition, memory cache keys have a code version (also called a build version) as part of the key (e.g., the code version may be appended to the beginning or end of the key). The code version is part of the key so that the keys-value pairs will be kept separate when code versions are dated. According to a second precondition, incoming web application requests have their build versions checked, and the web application server will force the client (e.g., a web browser client) to reload the client code if the build version of the client code is different from the build version of the server code. In this way, the requirement that the server and client are both running the same build version is enforced.
In addition to the preconditions, the database environment is setup as follows. There is one master database (a master copy of the database) and at least one slave database (at least one slave copy of the database). For example, there can be a master database, a first slave database (called slave_1) and a second slave database (called slave_2). There is a virtual IP address that can be switched between the master and slave databases. Having a virtual IP address provides advantages. For example, having a virtual IP address can reduce or eliminate the time window where some existing application server requests will write to the master database while new processes start writing to the slave database. Also, services do not need to be started.
In the example implementation, a number of operations are performed as a first step, as follows. The build version is updated to a new version, called version x+1, to the nodes in the environment (e.g., to the web application servers). Replication is disconnected between the master and slave. Changes to the slave database are performed, including schema changes associated with the new version x+1. The web application continues running and writing new changes to the master database at this time. There are some restrictions on the database schema changes, specifically new columns and fields must be added to the end of the table, and all new columns must have default values. The current build version of the application code must be able to run with the database schema changes (e.g., the new default values). Population of any new columns with non-default values must be done after the update process is complete. If columns need to be deleted from the database, it can be done in a subsequent build (build x+2) when the code that depends on it has been removed.
In a second step, replication is re-enabled from the master to the slave database. All changed rows in the master database since the replication was suspended are propagated to the slave database.
A third step begins once the slave database catches up with the master database (once they are in sync). At this time, any cron processes (processes that run on a periodic scheduled basis) and worker processes (processes that perform background processing tasks) are stopped. Any final changes to the database performed by the cron and worker processes are replicated to the slave database. Next, the slave database is re-labeled as the new master and the master database is re-labeled as the new slave. Note that replication is still running from the new slave to the new master. Next read-only is turned off on the new master. The virtual IP address is switched to point to the new master which is now in read-write mode. Replication is stopped from the new slave to the new master. The new slave is set to read-only mode. The worker and cron processes are started. Finally, replication is started from the new master to the new slave. If additional slaves are present, replication is started from the new master to the additional slaves.
In a final step, the new build version is activated on the web application servers. Any clients that connect to the web application servers that do not have the new build version will automatically download and use the new version.
With reference to
A computing system may have additional features. For example, the computing system 500 includes storage 540, one or more input devices 550, one or more output devices 560, and one or more communication connections 570. An interconnection mechanism (not shown) such as a bus, controller, or network interconnects the components of the computing system 500. Typically, operating system software (not shown) provides an operating environment for other software executing in the computing system 500, and coordinates activities of the components of the computing system 500.
The tangible storage 540 may be removable or non-removable, and includes magnetic disks, magnetic tapes or cassettes, CD-ROMs, DVDs, or any other medium which can be used to store information in a non-transitory way and which can be accessed within the computing system 500. The storage 540 stores instructions for the software 580 implementing one or more innovations described herein.
The input device(s) 550 may be a touch input device such as a keyboard, mouse, pen, or trackball, a voice input device, a scanning device, or another device that provides input to the computing system 500. For video encoding, the input device(s) 550 may be a camera, video card, TV tuner card, or similar device that accepts video input in analog or digital form, or a CD-ROM or CD-RW that reads video samples into the computing system 500. The output device(s) 560 may be a display, printer, speaker, CD-writer, or another device that provides output from the computing system 500.
The communication connection(s) 570 enable communication over a communication medium to another computing entity. The communication medium conveys information such as computer-executable instructions, audio or video input or output, or other data in a modulated data signal. A modulated data signal is a signal that has one or more of its characteristics set or changed in such a manner as to encode information in the signal. By way of example, and not limitation, communication media can use an electrical, optical, RF, or other carrier.
The innovations can be described in the general context of computer-executable instructions, such as those included in program modules, being executed in a computing system on a target real or virtual processor. Generally, program modules include routines, programs, libraries, objects, classes, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The functionality of the program modules may be combined or split between program modules as desired in various embodiments. Computer-executable instructions for program modules may be executed within a local or distributed computing system.
The terms “system” and “device” are used interchangeably herein. Unless the context clearly indicates otherwise, neither term implies any limitation on a type of computing system or computing device. In general, a computing system or computing device can be local or distributed, and can include any combination of special-purpose hardware and/or general-purpose hardware with software implementing the functionality described herein.
For the sake of presentation, the detailed description uses terms like “determine” and “use” to describe computer operations in a computing system. These terms are high-level abstractions for operations performed by a computer, and should not be confused with acts performed by a human being. The actual computer operations corresponding to these terms vary depending on implementation.
The cloud computing services 610 are utilized by various types of computing devices (e.g., client computing devices), such as computing devices 620, 622, and 624. For example, the computing devices (e.g., 620, 622, and 624) can be computers (e.g., desktop or laptop computers), mobile devices (e.g., tablet computers or smart phones), or other types of computing devices. For example, the computing devices (e.g., 620, 622, and 624) can utilize the cloud computing services 610 to perform computing operators (e.g., data processing, data storage, and the like).
Although the operations of some of the disclosed methods are described in a particular, sequential order for convenient presentation, it should be understood that this manner of description encompasses rearrangement, unless a particular ordering is required by specific language set forth below. For example, operations described sequentially may in some cases be rearranged or performed concurrently. Moreover, for the sake of simplicity, the attached figures may not show the various ways in which the disclosed methods can be used in conjunction with other methods.
Any of the disclosed methods can be implemented as computer-executable instructions or a computer program product stored on one or more computer-readable storage media and executed on a computing device (e.g., any available computing device, including smart phones or other mobile devices that include computing hardware). Computer-readable storage media are any available tangible media that can be accessed within a computing environment (e.g., one or more optical media discs such as DVD or CD, volatile memory components (such as DRAM or SRAM), or nonvolatile memory components (such as flash memory or hard drives)). By way of example and with reference to
Any of the computer-executable instructions for implementing the disclosed techniques as well as any data created and used during implementation of the disclosed embodiments can be stored on one or more computer-readable storage media. The computer-executable instructions can be part of, for example, a dedicated software application or a software application that is accessed or downloaded via a web browser or other software application (such as a remote computing application). Such software can be executed, for example, on a single local computer (e.g., any suitable commercially available computer) or in a network environment (e.g., via the Internet, a wide-area network, a local-area network, a client-server network (such as a cloud computing network), or other such network) using one or more network computers.
For clarity, only certain selected aspects of the software-based implementations are described. Other details that are well known in the art are omitted. For example, it should be understood that the disclosed technology is not limited to any specific computer language or program. For instance, the disclosed technology can be implemented by software written in C++, Java, Perl, JavaScript, Adobe Flash, or any other suitable programming language. Likewise, the disclosed technology is not limited to any particular computer or type of hardware. Certain details of suitable computers and hardware are well known and need not be set forth in detail in this disclosure.
Furthermore, any of the software-based embodiments (comprising, for example, computer-executable instructions for causing a computer to perform any of the disclosed methods) can be uploaded, downloaded, or remotely accessed through a suitable communication means. Such suitable communication means include, for example, the Internet, the World Wide Web, an intranet, software applications, cable (including fiber optic cable), magnetic communications, electromagnetic communications (including RF, microwave, and infrared communications), electronic communications, or other such communication means.
The disclosed methods, apparatus, and systems should not be construed as limiting in any way. Instead, the present disclosure is directed toward all novel and nonobvious features and aspects of the various disclosed embodiments, alone and in various combinations and sub combinations with one another. The disclosed methods, apparatus, and systems are not limited to any specific aspect or feature or combination thereof, nor do the disclosed embodiments require that any one or more specific advantages be present or problems be solved.
The technologies from any example can be combined with the technologies described in any one or more of the other examples. In view of the many possible embodiments to which the principles of the disclosed technology may be applied, it should be recognized that the illustrated embodiments are examples of the disclosed technology and should not be taken as a limitation on the scope of the disclosed technology. Rather, the scope of the disclosed technology includes what is covered by the scope and spirit of the following claims.
