Managed PostgreSQL autoscaling replicas

[matmicro]

As a Managed PostgreSQL user, i need :

• PgPool or equivalent integrated in order to split ReadWrite queries to the master and ReadOnly queries onto all replicas

• Database to automatically scale up by adding more read replicas nodes if CPU goes over a defined value (eg. 50%) and automatically remove extra nodes as soon as CPU becomes lower

• Also, it would be great to be able to switch to lower flavors, eg. change from DB1-120 to DB1-60

[matmicro]

@mhurtrel any chance to implement this as it is available on Aiven ?

[mhurtrel]

A question for my colleague @baaastijn ;)

[ZuSe]

I would also love to see that just disk space scales while keeping CPU/RAM. For us this has been one of the most used features on IBM Cloud and Azure in the past.

https://cloud.ibm.com/docs/databases-for-postgresql?topic=databases-for-postgresql-autoscaling

I am pretty sure Aiven is already able to offer it under the hood, just there is no configuration to set our desired values.

[baaastijn]

Hello @matmicro and @ZuSe, sorry i missed this issue.

Here is the situation.

Today, on PostgreSQL offers, we allow cluster creation, with one, two and three nodes (1 primary + 2 x replicas). Today you cannot add other replica additionally.

While we see a strong interest about adding replicas, almost 99% of the time it's also linked with Multi-Az : having these replicas in another physical location (for example 2 nodes in GRA and 2 nodes in FRA).

That's why we will focus on the roadmap this way :

1. work on more flexible storage (with option like standard and high perf, and dynamic resizing). This one is already engaged.
2. work on Multi AZ for all clusters, not only PostgreSQL
3. provide the ability to add new replicas in postgreSQL
4. Work on autoscaling triggers based on CPU/RAM/...

[matmicro]

@baaastijn great news thanks.

Please share ETA for each item as soon as you can. We need all those features asap :)