Good Solution
Our mysql based saas plattform has its peak periods in november and december.
We realised in the middle of the year that we would need to integrate read-write splitting in order to be able to cope with the expected traffic on our systems without totally oversizing our MySQL aurora setup. The task turned out to be not as easy and a quick turn around in achieving this would not be possible.
Heimdall enabled read write splitting and thus enabled us to downscale the aurora instance sizes by 50% and at the same time we had the "elasticity" in the reader databases scaling in and out to cover peak periods.
The added benefit came in the query caching with elastic ache redis which drastically speeded up the responsiveness of the system. This was a real Bonus of the Heimdall System.
We now have 2 Heimdall Proxies managed by a worker (controller).
Analysing the cost benefits of the system we can break it down as follows:
- Decreased aurora cost by nearly 40%
- Increased cost of the Heimdall cluster, network loadbalancer, extra redis cache
I would say we are now running at 15% less total cost but have the huge burden of read write splitting solved, plus the advantages gained of the query cache in Redis.
The company also provides excellent support as the setup especially when you go to the clustermode to achieve high availability, is not that easy. They helped us all the way! You really feel that they are experts in their domain.
Overall – I highly recommend the software if you need a quick fix to read write splitting.