An application initiates a data update (including insert, delete, and modify operations) request. After completing the update operation, the Master sends a response to the application immediately, and then replicates the data to the Slave. During the asynchronous replication, the Master does not need to wait for a response from the Slave. Therefore, the DB instance replicated in an asynchronous way often has a higher performance. However, since the data is not synchronized to the Slave in real time, if the Master fails when a latency occurs on the Slave, data may be inconsistent between the Master and Slave.
A full backup automatically created for a DB instance by RDS. Users can set the automated backup start time and backup retention period.
During a primary/standby switchover, the switchover is performed even if data is inconsistent between the primary and standby DB instances and the synchronization delay is no more than 5 minutes, thereby ensuring service availability. If the delay is longer than five minutes, the system does not perform the primary/standby switchover and stop database services to prevent data loss.
A company or an organization that provides cloud computing services.
A cold backup is performed when a system is stopped or being maintained. The backup data is completely the same as the data in the system at the point in time.
In the primary/standby HA architecture, data will be replicated to a standby DB instance from a primary after data is committed to the primary DB instance. Data replications come in three types: forced synchronous replication, semi-synchronous replication, and asynchronous replication.
A person who is responsible for managing databases. A DBA uses dedicated software to store and organize data. Their responsibilities include but are not limited to capacity planning, installation, configuration, database design, migration, performance monitoring, security, troubleshooting, and data backup and restoration.
A database master user account is different from the user cloud account and used only within the RDS instance environment to control access to users' DB instances. The database master user account is a native database user account used to connect to DB instances. For example, when creating a MySQL DB instance, root is the master user account by default and users can set the root password. After a DB instance is created, users can connect to the database using the database master user account. Subsequently, users can also create additional database user accounts to meet service requirements.
As services develop, a database needs to be migrated from an environment to another, for example, from a local data center to a cloud, or from a cloud to another cloud.
The underlying storage resources used for storing data and logs of a database permanently.
Database types are classified into relational databases and non-relational databases.
A DB engine is a core service for storing, processing, and protecting data. It can be used to control access permissions and process transactions rapidly to meet enterprise requirements. Every DB instance supports DB engines.
A DB instance is an isolated database environment in the cloud. It is a basic building block of RDS. A DB instance can contain multiple databases created by users and can be accessed using the same client tool and application as those used for accessing an isolated DB instance.
The DB instance class determines the computing and memory capacity of a DB instance. A user can change the CPU or memory of an available DB instance by changing its DB instance class.
Each DB instance has a DB instance ID. This ID uniquely identifies a DB instance when a user uses the RDS console or RDS APIs. The DB instance ID must be unique for a user in a region.
A DB instance lifecycle starts from the time when the DB instance is created to the time when the DB instance is deleted. During a DB instance lifecycle, users can back up, restore, change instance classes, scale up storage space, reboot, or delete the instance.
A database parameter group functions as a container for engine configuration values that can be applied to one or more DB instances. If users create a DB instance without specifying a DB parameter group, the default parameter group is used. The default parameter group contains the default values of the engine and database system optimized for the running DB instances. If users want their DB instances to run with their self-defined engine configuration values, they can simply create a new database parameter group, modify certain parameters, and associate the new DB parameter group to a DB instance. Once associated, all DB instances that use this particular DB parameter group get all the parameter updates to that DB parameter group.
If an unexpected interruption occurs on a primary DB instance, RDS automatically switches to the standby DB instance to restore database operations quickly without intervention. The time required for completing a failover depends on the database activity and other conditions at the time the primary DB instance became unavailable. The failover time ranges from seconds to minutes. However, large transactions or lengthy recovery processes may increase the failover time.
For MySQL or PostgreSQL HA DB instances, when the switchover policy priority is availability first, the primary DB instance may have more data than the standby instance due to synchronization delay or other reasons before a switchover. After a switchover occurs, the primary DB instance is demoted to be standby and the standby DB instance is promoted to be primary. The data that has not be synchronized to the new primary DB instance (original standby) will be packaged and uploaded to OBS in SQL statements for users to download them.
A system availability that keeps a service running properly without interruption.
A hot backup is performed when a system is properly running. The backup data may be different from the actual data of the system because the data in the system keeps being updated.
RDS automatically backs up data updated after the last automated or incremental backup every five minutes.
The number of I/O operations completed per second. This metric is reported as the average IOPS for a given time interval. RDS reports read and write IOPS separately at one minute intervals. Total IOPS is the sum of the read and write IOPS. Typical values for IOPS range from zero to tens of thousands per second.
License type associated with a DB engine.
A procedure in which the structured query language (SQL) is used to abstract data from a database and store the data into a binary file. Logical backup is a technology that uses software to export data from a database and stores the data into a file which is in a format different from the file in the original database. Logical backup can only be used for logical restoration (data import) and cannot be used for physical restoration based on storage characters of the original database. Generally, logical backup is used for incremental backup and only backs up data that has changed since the last backup.
Network transmit throughput: indicates the rate of network traffic to and from the DB instance in MB per second.
Data in a NoSQL Database is communicated and organized based on the non-relational data structure. Based on different data structures, NoSQL Database can be classified as the following types: key-value, column-oriented, document-oriented, and graphic databases. Common NoSQL databases include: Memcached, Redis, MongoDB, Cassandra, HBase, MemacheDB, and BerkeleyDB.
The number of client sessions that are connected to the DB instance.
Reflect metrics of DB instance performance, including CPU usage, memory usage, storage space utilization, network traffic, database connections, transaction rate/database throughput, submit latency, storage latency, storage IOPS, storage throughput, and storage queue length.
A backup mechanism, in which database files in the operating system are replicated from one place to another place, generally from a disk to a tape. Physical backup includes cold backup and hot backup.
A DB instance that provides read and write services.
The number of I/O requests in the queue waiting to be serviced. These are I/O requests that have been submitted by the application but have not been sent to the device because the device is busy servicing other I/O requests. Time spent waiting in the queue is a component of Latency and Service Time (not available as a metric). This metric is reported as the average queue depth for a given time interval. RDS reports queue depth at one minute intervals. Typical values for queue depth range from zero to several hundred.
An RDS DB instance is the minimum RDS management unit. An RDS DB instance represents a relational database that runs independently. Users can create and manage DB instances of various database engines in the RDS system. RDS DB instances come in three types: primary DB instances, standby DB instances, and read replicas.
An active copy of another DB instance. Any updates to the data on the source DB instance are replicated to the read replica DB instance using the built-in replication feature of DB Engine.
Enable the master instance to handle INSERT, UPDATE, and DELETE operations while the slave instance to handle SELECT operations.
Tables in an RDB are communicated and organized based on the relational data structure. RDB simplifies complex data structures into simple binary relation (two-dimensional tables). A relational database contains multiple tables and each table is known as a relation. Data management is performed through data manipulation languages (DMLs) GROUP BY, JOIN, UNION, and SELECT * FROM. Common relational databases include: Oracle, MySQL, MariaDB, Microsoft SQL Server, Access, DB2, PostgreSQL, Informix, and Sybase.
RDS is a managed service that makes it easy to create, configure, operate, and scale a relational database in the cloud.
During a primary/standby switchover, if data is inconsistent between the primary and standby DB instances, the switchover is not performed and the database stops providing services. The reliability priority policy ensures data consistency.
A security protocol that works at a socket layer. This layer exists between the TCP layer and the application layer to encrypt/decode data and authenticate concerned entities.
An application initiates a data update (including insert, delete, and modify operations) request. After completing the update operation, the Master replicates data to a Slave. When at least one Slave receives the binlog, writes it to relay-log, and flushes it to the disk, the Slave can return a response to the Master. Compared to strong synchronous replication, semi-synchronous improves data replication performance because the Master does not wait for the Slave to flush the binlog to the disk. However, since the Slave responds to the Master before the commitment is done, data may be inconsistent between the Master and Slave.
SSDs are built on solid electronic storage chip arrays. Each SSD consists of a control unit and a storage unit (a flash and a DRAM chip). The interface specifications, definition, functions, and usage of an SSD are the same as those of a common hard disk. SSDs are widely applied in fields such as military, vehicles, industrial control, video surveillance, network surveillance, network terminals, electricity, medical, aeronautics, and navigation equipment.
A source DB instance functions as the data source in data replication.
A standby DB instance is a backup for the primary DB instance. It automatically takes over services from the standby DB instance in case of failures to enhance database availability. When creating a primary DB instance, users can determine whether to create a synchronous standby DB instance with the same specifications as the primary one.
The space of underlying storage resources for storing data and logs of a database.
Storage can be classified as different types based on the dimensions and attributes of storage resources. For example, storage can be classified into magnetic storage and SSD based on storage media, and can be classified into general, medium, and high performance I/O based on the I/O level.
An application initiates a data update (including insert, delete, and modify operations) request. After completing the update operation, the Master replicates data to a Slave immediately. After receiving the data, the Slave returns a success message to the Master. Only after receiving a message from the Slave, the Master can return a response to the application. Since data is replicated synchronously from the Master to the Slave, unavailability of the Slave will affect the operations on the Master, and unavailability of the Master will not cause data inconsistency.
The elapsed time between the submission of a request and its completion. This metric is closely related with the storage write latency metric. A high storage write latency may cause a high submit latency.
Policy used to switch over the primary DB instance to the standby DB instance in the primary/standby HA architecture. The switchover policy priority can be reliability or availability. By default, reliability is selected.
The number of bytes per second transferred to or from a disk. This metric is reported as the average throughput for a given time interval. RDS reports read and write throughput separately at one minute intervals using units of megabytes per second (MB/s). Typical values for throughput range from zero to the I/O channel's maximum bandwidth.
Number of completed transactions in a specified period, generally expressed in transactions per minute (TPM) or transactions per second (TPS). Another term of Transaction Rate is Database Throughput. Do not confuse it with the disk throughput. They may be irrelevant. Databases achieving a high transaction rate may have little or disk throughput, for example, by reducing load by reading from their cache.
Thank you for your score！Your feedback would help us improve the website.