All specifications listed below are per shard. For cluster and read/write splitting instances, multiply per-shard values by the number of shards to get total instance capacity.
The ESSD cloud disk included in each shard is reserved for system operations — storing logs, temporary backup files, and append-only files (AOF). It is not used for data storage.
Instance types
Standard architecture
The following specifications apply to cloud native instances that use standard (non-cluster) architecture.
If read/write splitting is enabled on a standard architecture instance, use the proxy mode and read/write splitting specifications instead.ESSD (GB)
| Instance type | Instance type code | CPU cores | Memory (GB) | Bandwidth | Maximum connections | Reference QPS | |
|---|---|---|---|---|---|---|---|
| 256 MB | redis.shard.micro.ce | 3 | 0.25 | 1 | 192 Mbps (24 MB/s) | 10,000 | 100,000 |
| 1 GB | redis.shard.small.2.ce | 3 | 1 | 5 | 384 Mbps (48 MB/s) | 10,000 | 100,000 |
| 2 GB | redis.shard.mid.2.ce | 3 | 2 | 10 | 768 Mbps (96 MB/s) | 10,000 | 100,000 |
| 4 GB | redis.shard.large.ce | 3 | 4 | 20 | 768 Mbps (96 MB/s) | 20,000 | 100,000 |
| 8 GB | redis.shard.xlarge.ce | 3 | 8 | 40 | 768 Mbps (96 MB/s) | 20,000 | 100,000 |
| 16 GB | redis.shard.2xlarge.ce | 3 | 16 | 80 | 768 Mbps (96 MB/s) | 20,000 | 100,000 |
| 24 GB | redis.shard.3xlarge.ce | 3 | 24 | 120 | 768 Mbps (96 MB/s) | 30,000 | 100,000 |
| 32 GB | redis.shard.4xlarge.ce | 3 | 32 | 160 | 768 Mbps (96 MB/s) | 30,000 | 100,000 |
| 64 GB | redis.shard.8xlarge.ce | 3 | 64 | 320 | 768 Mbps (96 MB/s) | 30,000 | 100,000 |
Cluster architecture — direct connection mode
The following specifications apply to cloud native instances that use cluster architecture in direct connection mode. For details on direct connection mode, see Cluster architecture.
ESSD (GB)| Instance type | Instance type code | CPU cores | Memory (GB) | Bandwidth | Maximum connections | Reference QPS | |
|---|---|---|---|---|---|---|---|
| 1 GB | redis.shard.small.ce | 3 | 1 | 5 | 384 Mbps (48 MB/s) | 10,000 | 100,000 |
| 2 GB | redis.shard.mid.ce | 3 | 2 | 10 | 768 Mbps (96 MB/s) | 10,000 | 100,000 |
| 4 GB | redis.shard.large.ce | 3 | 4 | 20 | 768 Mbps (96 MB/s) | 20,000 | 100,000 |
| 8 GB | redis.shard.xlarge.ce | 3 | 8 | 40 | 768 Mbps (96 MB/s) | 20,000 | 100,000 |
| 16 GB | redis.shard.2xlarge.ce | 3 | 16 | 80 | 768 Mbps (96 MB/s) | 20,000 | 100,000 |
| 24 GB | redis.shard.3xlarge.ce | 3 | 24 | 120 | 768 Mbps (96 MB/s) | 30,000 | 100,000 |
| 32 GB | redis.shard.4xlarge.ce | 3 | 32 | 160 | 768 Mbps (96 MB/s) | 30,000 | 100,000 |
| 64 GB | redis.shard.8xlarge.ce | 3 | 64 | 320 | 768 Mbps (96 MB/s) | 30,000 | 100,000 |
Cluster architecture — proxy mode and read/write splitting architecture
The following specifications apply to cloud native instances that use cluster architecture in proxy mode or the read/write splitting architecture. For details on proxy mode, see Cluster architecture.
ESSD (GB)| Instance type | Instance type code | CPU cores | Memory (GB) | Bandwidth | Maximum connections | Reference QPS | |
|---|---|---|---|---|---|---|---|
| 1 GB | redis.shard.with.proxy.small.ce | 3 | 1 | 5 | 384 Mbps (48 MB/s) | 10,000 | 100,000 |
| 2 GB | redis.shard.with.proxy.mid.ce | 3 | 2 | 10 | 768 Mbps (96 MB/s) | 10,000 | 100,000 |
| 4 GB | redis.shard.with.proxy.large.ce | 3 | 4 | 20 | 768 Mbps (96 MB/s) | 20,000 | 100,000 |
| 8 GB | redis.shard.with.proxy.xlarge.ce | 3 | 8 | 40 | 768 Mbps (96 MB/s) | 20,000 | 100,000 |
| 16 GB | redis.shard.with.proxy.2xlarge.ce | 3 | 16 | 80 | 768 Mbps (96 MB/s) | 20,000 | 100,000 |
| 24 GB | redis.shard.with.proxy.3xlarge.ce | 3 | 24 | 120 | 768 Mbps (96 MB/s) | 30,000 | 100,000 |
| 32 GB | redis.shard.with.proxy.4xlarge.ce | 3 | 32 | 160 | 768 Mbps (96 MB/s) | 30,000 | 100,000 |
| 64 GB | redis.shard.with.proxy.8xlarge.ce | 3 | 64 | 320 | 768 Mbps (96 MB/s) | 30,000 | 100,000 |
Cloud native instances in cluster architecture support adjustable shard counts. To change the number of shards after creation, see Adjust the number of shards for a cluster instance.
CPU cores
Each shard has 3 CPU cores:
-
2 working cores — handle user requests
-
1 background core — handles non-service tasks: persistence, accelerated expiration, performance analysis, and logging
Bandwidth
-
All bandwidth values in the tables are per shard. Total bandwidth for cluster and read/write splitting instances equals the sum across all shards.
-
Bandwidth applies independently to upstream and downstream traffic. For example, an instance with 192 Mbps bandwidth has 192 Mbps upstream and 192 Mbps downstream simultaneously.
-
The bandwidth limit applies to shard nodes and is independent of the network connection type. This applies to both Tair and Redis Open-Source Edition instances.
-
Proxy mode and read/write splitting: Total bandwidth is capped at 16,384 Mbps (2,048 MB/s). Adding more shards does not increase bandwidth beyond this limit.
-
Direct connection mode: No overall bandwidth cap.
For more information, see Bandwidth FAQ.
Maximum connections
-
Proxy mode: Maximum connections per instance is 500,000. This cap does not increase with additional shards or nodes.
-
Direct connection mode: Maximum connections per shard is 10,000. Total instance connections = number of shards x 10,000.
FAQ
Do different instance types have different performance?
The 256 MB and 1 GB instances have lower bandwidth than the 768 Mbps (96 MB/s) available on larger types. All instances of 2 GB and larger share the same CPU and bandwidth specifications — the only difference is memory capacity.