This page defines the key terms for ApsaraDB for ClickHouse. Review these concepts before configuring your cluster.
Common terms
Region
A region is the physical location where ApsaraDB for ClickHouse servers are hosted. Select a region when you purchase the service. The region cannot be changed after purchase.
Zone
A zone is a physical area within a region with an independent power supply and network. Zones in the same region are connected through a low-latency internal network.
Database
A database is the highest-level object in an ApsaraDB for ClickHouse cluster. It contains objects such as tables, columns, views, functions, and data types.
Community-compatible Edition
ApsaraDB for ClickHouse cluster
An ApsaraDB for ClickHouse cluster is a distributed database consisting of multiple ClickHouse Server instances. Depending on the specifications you purchase, the cluster may contain one or more shards and one or more replicas. Logically, a cluster can contain multiple database objects.
Edition
The Edition determines the number of replicas per shard and affects availability and cost.
Double-replica Edition: Each shard contains two replicas. If one replica becomes unavailable, the other continues to serve traffic. Data is replicated across both replicas to make sure consistency is maintained.
ImportantWhen creating tables in a Double-replica Edition cluster, use Replicated table engines from the MergeTree family. Non-replicated table engines do not replicate data between replicas, which can cause data inconsistency.
Single-replica Edition: Each shard has one replica. If the replica becomes unavailable, the cluster becomes unavailable until the replica is fully restored.
A Double-replica Edition cluster uses twice the resources and costs twice as much as a Single-replica Edition cluster. Because the underlying disks provide high reliability, a Single-replica Edition cluster still prevents data loss.
Shard
A shard is a server that stores and processes a portion of the total data in a cluster. When the storage or compute capacity of a single server becomes a bottleneck, ApsaraDB for ClickHouse distributes data across multiple shards to improve efficiency.
Replica
A replica is a redundant copy of data stored on two or more servers. ApsaraDB for ClickHouse uses replication to protect against server failures and maintain high availability.
Table
A table is the basic structure for storing data. It consists of rows and columns, where each column represents a field and each row represents a record.
By data distribution:
| Table type | Description |
|---|---|
| Local table | Stores data only on the node to which it is written. Writes and queries are limited to a single server's resources and do not scale horizontally. |
| Distributed table | A logical view over a collection of local tables. Writes are automatically distributed across the underlying local tables; queries are executed against each local table and the results are aggregated. Supports horizontal scaling. |
By storage engine:
| Table type | Description |
|---|---|
| Non-replicated table | Stores data only on the current server with a single replica. Cannot guarantee high availability if the server fails. |
| Replicated table | Automatically replicates data to multiple servers. Can serve traffic as long as at least one replica is available. |
Data part
A data part is a self-contained fragment of data stored on disk. It is the basic unit of storage for a ClickHouse table. Each time data is written to a ClickHouse table, a new data part is generated. Each data part includes all columns and indexes for its portion of the data and maintains data order. This design enables efficient merge and compression operations for high-performance query processing.
Enterprise Edition
ApsaraDB for ClickHouse cluster
An ApsaraDB for ClickHouse cluster in the Enterprise Edition is composed of computing and storage resource units. It provides a Platform as a Service (PaaS) for data storage and analysis based on the ClickHouse engine.
Worker node
A worker node is a replica node within an ApsaraDB for ClickHouse cluster. It is the physical resource that participates in engine computations.
CCU
A ClickHouse Compute Unit (CCU) is the unit for measuring and billing computing resources in an ApsaraDB for ClickHouse cluster. One CCU equals 1 vCPU and 4 GiB of memory. The standard billing unit is CCU per minute.
Auto scaling of computing resources
Auto scaling of computing resources automatically adjusts the number of CCUs in a cluster based on CPU and memory usage.
Auto scaling range
The auto scaling range defines the minimum and maximum number of CCUs for a cluster. The auto scaling feature keeps the CCU count within this range.
Storage resources
Storage resources refer to the shared storage solution used by the Enterprise Edition. They are billed on a pay-as-you-go basis.