This topic describes the types and parameters of all ApsaraDB for Redis editions and the method of testing queries per second (QPS). When calling APIs related to instance types, you may need to queryInstanceClassParameters.

Note
  • The value of intranet bandwidth applies to both upstream and downstream. If network resources are sufficient, the bandwidth is unlimited for ApsaraDB for Redis instances. However, if network resources are insufficient, the intranet bandwidth takes effect for the instances.
  • The intranet bandwidth of cluster instances = the number of nodes × the intranet bandwidth of a single node (96 MB/s ). The maximum intranet bandwidth of an instance is 2048 MB/s.
  • The maximum concorrent connections of cluster instances = the number of nodes × the maximum concorrent connections per node (10000 ). The maximum number of maximum concorrent connections is 500000.

Standard dual-replica edition

Size (GB) InstanceClass (for calling APIs) New connections per second Maximum concorrent connections Intranet bandwidth(MB/s) CPU processing capability QPS reference value Description
1 GB master/replica redis.master.small.default 10,000 10,000 10 Single-core 80,000 Master-replica instance
2 GB master/replica redis.master.mid.default 10,000 10,000 16 Single-core 80,000 Master-replica instance
4 GB master/replica redis.master.stand.default 10,000 10,000 24 Single-core 80,000 Master-replica instance
8 GB master/replica redis.master.large.default 10,000 10,000 24 Single-core 80,000 Master-replica instance
16 GB master/replica redis.master.2xlarge.default 10,000 10,000 32 Single-core 80,000 Master-replica instance
16 GB master/replica redis.master.4xlarge.default 10,000 10,000 32 Single-core 80,000 Master-replica instance
64 GB master/replica redis.master.8xlarge.default 10,000 10,000 48 Single core 80,000 Master-replica instance

Standard zone-disaster recovery edition

Note To create a zone-disaster recovery instance, you have to select the region and zone that support zone-disaster recovery, such asChina (Hangzhou) Zone (G + H ).
Type InstanceClass (for calling APIs) New connections per second Maximum concorrent connections Intranet bandwidth(MB/s) CPU processing capability QPS reference value Description
1 GB zone-disaster recovery redis.logic.sharding.drredissdb1g.1db.0rodb.4proxy.default 10,000 10,000 10 Single-core 80,000 Master/replica zone-disaster recovery instances
2 GB zone-disaster recovery redis.logic.sharding.drredissdb2g.1db.0rodb.4proxy.default 10,000 10,000 16 Single-core 80,000 Master/replica zone-disaster recovery instances
4 GB zone-disaster recovery redis.logic.sharding.drredissdb4g.1db.0rodb.4proxy.default 10,000 10,000 24 Single-core 80,000 Master/replica zone-disaster recovery instances
8 GB zone-disaster recovery redis.logic.sharding.drredissdb8g.1db.0rodb.4proxy.default 10,000 10,000 24 Single-core 80,000 Master/replica zone-disaster recovery instances
16 GB zone-disaster recovery redis.logic.sharding.drredissdb16g.1db.0rodb.4proxy.default 10,000 10.000 32 Single-core 80,000 Master/replica zone-disaster recovery instances
32 GB zone-disaster recovery redis.logic.sharding.drredissdb32g.1db.0rodb.4proxy.default 10,000 10,000 32 Single-core 80,000 Master/replica zone-disaster recovery instances
64 GB zone-disaster recovery redis.logic.sharding.drredissdb64g.1db.0rodb.4proxy.default 10,000 10,000 48 Single-core 80,000 Master/replica zone-disaster recovery instances

Dual-replica cluster edition

Type InstanceClass (for calling APIs) Number of nodes New connections per second Maximum concorrent connections Intranet bandwidth(MB/s) CPU processing capability QPS reference value Description
16 GB cluster redis.logic.sharding.2g.8db.0rodb.8proxy.default 8 50,000 80,000 768 8-core 640,000 High-performance cluster instance
32 GB cluster redis.logic.sharding.4g.8db.0rodb.8proxy.default 8 50,000 80,000 768 8-core 640,000 High-performance cluster instance
64 GB cluster redis.logic.sharding.8g.8db.0rodb.8proxy.default 8 50,000 80,000 768 8-core 640,000 High-performance cluster instance
128 Gb cluster redis.logic.sharding.8g.16db.0rodb.16proxy.default 16 50,000 160,000 1,536 16-core 1,280,000 High-performance cluster instance
256 GB cluster redis.logic.sharding.16g.16db.0rodb.16proxy.default 16 50,000 160,000 1,536 16-core 1,280,000 High-performance cluster instance
512 GB cluster redis.logic.sharding.16g.32db.0rodb.32proxy.default 32 50,000 320,000 2,048 32-core 2,560,000 High-performance cluster instance
Note The number of nodes in the dual-replica cluster instance is the number of master nodes.

Zone-disaster recovery cluster edition

Note To create a zone-disaster recovery instance, you have to select the region and zone that support zone-disaster recovery, such asChina (Hangzhou) Zone (G + H ).
Type InstanceClass (for calling APIs) Number of nodes New connections per second Maximum concorrent connections Intranet bandwidth(MB/s) CPU processing capability QPS reference value Description
16 GB zone-disaster recovery cluster redis.logic.sharding.drredismdb16g.8db.0rodb.8proxy.default 8 50,000 80,000 768 8-core 800,000 Zone-disaster recovery cluster instance
32 GB zone-disaster recovery cluster redis.logic.sharding.drredismdb32g.8db.0rodb.8proxy.default 8 50,000 80,000 768 8-core 800,000 Zone-disaster recovery cluster instance
64 GB zone-disaster recovery cluster redis.logic.sharding.drredismdb64g.8db.0rodb.8proxy.default 8 50,000 80,000 768 8-core 800,000 Zone-disaster recovery cluster instance
128 GB zone-disaster recovery cluster redis.logic.sharding.drredismdb128g.16db.0rodb.16proxy.default 16 50,000 160,000 1,536 16-core 1,600,000 Zone-disaster recovery cluster instance
256 GB zone-disaster recovery cluster redis.logic.sharding.drredismdb256g.16db.0rodb.16proxy.default 16 50,000 160,000 1,536 16-core 1,600,000 Zone-disaster recovery cluster instance
512 GB zone-disaster recovery cluster redis.logic.sharding.drredismdb512g.32db.0rodb.32proxy.default 32 50,000 320,000 2,048 32-core 3,200,000 Zone-disaster recovery cluster instance

QPS performance reference

QPS performance

Type Maximum concorrent connections Intranet bandwidth(MB/s) CPU processing capability QPS reference value
8 GB 10,000 24 Single-core 80,000
Note

QPS reference values of non-cluster instances range from 80,000 to 100,000. QPS reference values of cluster instances are the number of nodes multiplied by the value from 80,000 to 100,000.

QPS testing method

Figure 1. Network topology

ECS instance type

Operating system CPU processing capability Memory Region Number of hosts
Ubuntu 14.04 64-bit 1-core 2048 MB China (Shenzhen) 3

Steps

  1. Download the source code package for redis-2.8.19 to three ECS servers.

     $ wget http://download.redis.io/releases/redis-2.8.19.tar.gz
     $ tar xzf redis-2.8.19.tar.gz
     $ cd redis-2.8.19
     $ make
     $ make install
  2. Run the following command on these ECS instances at the same time.

     redis-benchmark -h ***********.m.cnsza.kvstore.aliyuncs.com -p 6379 -a password -t set -c 50 -d 128 -n 25000000 -r 5000000
  3. Summarize the testing data on these ECS instances. The total QPS is the sum of the QPS on each ECS instance.