Read replica in Redshift ?

0

Hello

I have a dc2.large Redshift cluster that I'm using to run some ETL processes, and I'm also using Metabase to generate dashboards from a table. This seems to put a strain on the database, as even simple queries stop running.

I don't see any spikes in CPU or memory usage in my cluster dashboard, or any degradation in the cluster, so the problem is not that obvious

I'm considering creating a secondary cluster to use as a read replica, but I'm not sure if there's a native option other than materialized tables. Datashare isn't an option because it doesn't support dc2.large nodes, and DMS also isn't viable since there's no endpoint for Redshift, and the PostgreSQL one doesn't work either.

I asked ChatGPT for a solution, and it suggested using UNLOAD and LOAD with S3 as an intermediary, but I need near real-time synchronization between my clusters.

Is upgrading or adding nodes to my cluster the only way to achieve this?

Note: I don't have control over the ETL process, Metabase integration. nor databases/tables indexation

1 Answer
0
Accepted Answer

Hello, DC2 is the previous generation node type and on its way to being deprecated. I would strongly suggest you consider upgrading to the ra3.xlplus node type (the smallest node type at the moment in the RA3 family). With RA3, you can use Amazon Redshift's native data sharing capability to setup a read-replica and workload isolation with just a few steps in the Redshift console, and since it's data sharing - the data is queried in-place without any physical copying or loading. It is also possible that when you upgrade to the ra3.xlplus, the newer node type will be able to handle all your workloads and the need for a read-replica will go away. This is the simplest solution with best-practices in mind. In addition to the ra3.xlplus node type, you could consider Amazon Redshift serverless if you workload only runs a few hours in the day. Hope this helps. Let me know if any other questions.

AWS
asser_m
answered 15 days ago
  • It is more expensive that my current node, I will try to switch to serverless and monitor cost thanks !

You are not logged in. Log in to post an answer.

A good answer clearly answers the question and provides constructive feedback and encourages professional growth in the question asker.

Guidelines for Answering Questions