- Neueste
- Die meisten Stimmen
- Die meisten Kommentare
Brian,
Warming up the green instance in an RDS MySQL blue/green deployment typically involves running queries to load data into the instance's cache. Running SELECT * queries on large tables is indeed not practical. Here are some alternative strategies:
- Targeted Queries: Instead of SELECT *, run queries that target the most frequently accessed data or use LIMIT to reduce the amount of data loaded.
- Replication: Use the blue instance as the replication source to keep the green instance warm, ensuring the cache is populated with recent queries.
- Incremental Load: Gradually increase the workload on the green instance before the full cutover.
Currently, RDS does not have a feature equivalent to EC2's Fast Restore Snapshot that instantly warms the cache.
I'm here to help!
Thank you, Vitor.
Do you know if there's anything upcoming in the AWS roadmap to help deal with lazy load in a more sensible manner?
The ideal solution would be to have a button in the Console to force all of the data to be copied from S3 into the disk. It would take hours of course, but IRL there's no real alternative -- and a Console button would be way simpler than having to come up with ways to somehow warm up the data.
We ran multiple tests here and the added latency makes any real-life usage impossible. Queries that should take 200ms are taking 20 minutes. As it stands, the DB ends up being restored, but is useless for production. This is particularly bad in a blue/green environment approach.
Relevanter Inhalt
- AWS OFFICIALAktualisiert vor 3 Jahren
- AWS OFFICIALAktualisiert vor 2 Jahren
- AWS OFFICIALAktualisiert vor 3 Jahren