Common terms used in cloud-native database PolarDB.
Region
The geographic location of a data center.
Zone
A physical area within a region with independent power and network. Instances in the same zone have lower network latency.
Cluster
A PolarDB cluster contains one primary node and multiple read-only nodes. A PolarDB cluster can span zones but not regions.
Global Database Network
A network of multiple PolarDB clusters distributed across regions with synchronized, consistent data. Global Database Network (GDN).
Primary Cluster
The single cluster in a GDN with read and write permissions.
Secondary Cluster
A subordinate cluster in a GDN that synchronizes data from the primary cluster.
Node
A PolarDB cluster consists of primary nodes and read-only nodes. Nodes of the same type are peers with identical specifications.
Primary node
A PolarDB cluster has one primary node (read/write node).
Read-only node
A PolarDB cluster can have up to 15 read-only nodes.
Cluster zone
The two zones across which cluster data is replicated for disaster recovery. Nodes can migrate only between these two zones.
Primary zone
The zone where the primary node of a PolarDB cluster is located.
Failover
The process of promoting a read-only node to the primary node, also known as a primary/standby switch. Automatic or manual failover.
Specifications
The resource configuration of each node in a PolarDB cluster, such as 8 cores and 64 GB. Compute node specifications for Enterprise Edition.
Endpoint
An access point for connecting to a database. Each cluster provides multiple endpoints, each connecting to one or more nodes. The primary endpoint points to the primary node; the cluster endpoint connects to primary and read-only nodes for read/write splitting. Endpoint properties include read/write status, node list, load balancing, and consistency level.
Address
The representation of an endpoint on a specific network plane. An endpoint can have both a private and a public address. Address properties include domain name, IP address, VPC, and vSwitch.
Primary Endpoint
The endpoint of the primary node. If a failover occurs, the system automatically redirects the endpoint to the new primary node.
Cluster Endpoint
A unified address that integrates multiple nodes for read/write splitting, load balancing, automatic scaling, and consistency coordination. Can be configured as read-only or read/write. Connection endpoints (primary, cluster, and custom endpoints).
Eventual Consistency
The default consistency level for read-only mode, providing optimal performance for a PolarDB cluster. Eventual consistency.
Session Consistency
The default consistency level for read/write mode, also known as causal consistency. Provides session-level read consistency suitable for most application scenarios. Session consistency.
Global Consistency
The highest consistency level, also known as strong or cross-session consistency. Guarantees consistency across sessions but increases load on the primary node. Not suitable when replication delay is high. Global consistency.
Transaction Splitting
A cluster endpoint feature that routes SELECT queries within transactions to read-only nodes, reducing primary node load while maintaining consistency. Transaction splitting.
Offload Reads from Primary Node
A cluster endpoint feature that routes SQL queries to read-only nodes, reducing primary node load while maintaining consistency. What is read/write splitting?.
Private Domain Name
PolarDB integrates with PrivateZone to bind private domain names to internal addresses of PolarDB primary and cluster endpoints. Effective only within the specified VPC in the current region. Private domain names.
Snapshot Backup
The backup method for PolarDB data. Currently the only supported backup method. Backup and recovery.
Level-1 Backup
A backup stored in the distributed storage cluster. Offers the fastest backup and recovery at higher cost. Backup and recovery.
Level-2 Backup
A backup stored in offline media, sourced from level-1 backups. Can be permanently retained. Low cost with slower recovery. Backup and recovery.
Log Backup
A redo log backup used for point-in-time restore (PITR). Retained for at least 7 days to prevent data loss. Uses low-cost offline storage. Backup and recovery.
Storage Capacity
The space occupied by data files, index files, log files (online logs and archived logs), and temporary files.
A newly purchased PolarDB cluster automatically creates database operation files (including the types listed above), which occupy some storage space.
Cluster Edition
The recommended product series for PolarDB for MySQL. Cluster Edition uses a compute-storage decoupled architecture with 2 to 16 dynamically scalable nodes. Cluster Edition.
Smart-SSD
A hardware-level compression technology using dedicated FPGA/ASIC chips embedded in SSDs. Compresses and decompresses data in real time during reads and writes, reducing storage costs with near-standard disk performance. Optimized for PolarDB with standard disk interfaces, transparent to applications. Used in the PSL4 storage class of PolarDB.
Hot standby storage cluster
A storage cluster deployed in a secondary zone or different data center for hot-standby failover. Has independent storage capabilities. If the PolarDB cluster or primary zone becomes unavailable, the hot standby cluster is promoted to handle read and write operations. Disabling the hot standby storage cluster reduces the PolarDB cluster SLA.