Benchmark – High-Availability Read-Replicas

Benchmark statistics are based on product replication engine version 6.0 test results.
Product versions 9.0 and above deliver substantially better performance.

AWS High-Availability Read-Replicas Scenario

High-Availability Read-Replicas Scenario:

Inserting 100 records per table every 10 seconds, first into 4 tables (in a 54 tables database). And then increasing the load to inserting 100 records per table every 10 seconds into all 54 tables in the database. The database is hosted in AWS Virginia based RDS db.M4.2Xlarge instance, replicating to another zone (Multi-AZ with creating of a Readable-Replica) asynchronously to a smaller RDS db.M4.Large server with SSD2 drives. Replacing the SSD2 drives with SSD drives with 1800 or above IOPS is expected to yield even better results.

Number of tables in the Database: 54

Volume of Changes: ~100 records per table inserted concurrently first into 4 tables, and later into all 54 tables

[Replication lag: 2 - 22 seconds]

Source RDS Type: db.M4.2Xlarge
CLOUDBASIC RDS AlwaysOn/Geo-Replicate EC2 type: ec2.M4.XLarge
Replica RDS Type: db.M4.large


RDS AlwaysOn/Geo-Replicate Dashboard: replication load inreases from ~100 concurrent inserts into 4 tables to ~100 concurrent inserts into 54 tables. The lag increases from ~2 seconds to ~22 seconds.


RDS AlwaysOn/Geo-Replicate Dashboard: ~100 pending changes per table at any given moment
Replication lag: <22 seconds


Virginia RDS db.M4.2XLarge (Primary: ~100 concurrent inserts into 54 tables every 10 seconds)

Resource Utilization Statistics


Virginia RDS db.M4.Large (Read-Replica: ~100 concurrent inserts into 54 tables every 10 seconds with <20 seconds lag)

Resource Utilization Statistics

Oregon RDS R3.Large


<< Back to Benchmarks