Many rely on multi-AZ setups, but still face bottlenecks when it comes to write performance.
Here's a little-known secret: bumping your max_wal_size from 2GB to 20GB can significantly speed up inserts.
But beware...
Many rely on multi-AZ setups, but still face bottlenecks when it comes to write performance.
Here's a little-known secret: bumping your max_wal_size from 2GB to 20GB can significantly speed up inserts.
But beware...
Your database can handle queries - until it can't.
PostgreSQL starts choking at 100-300 connections.
When a failover hits, your app gets slammed with reconnect requests.
RDS Proxy keeps connections alive, so...
calvintorra.com/hire-me/rds-...
Your database can handle queries - until it can't.
PostgreSQL starts choking at 100-300 connections.
When a failover hits, your app gets slammed with reconnect requests.
RDS Proxy keeps connections alive, so...
calvintorra.com/hire-me/rds-...
Failover in AWS RDS takes ~70s, but full recovery can stretch to 4 minutes due to a cold cache.
- After a restart, shared buffers are empty, slowing queries.....
Failover in AWS RDS takes ~70s, but full recovery can stretch to 4 minutes due to a cold cache.
- After a restart, shared buffers are empty, slowing queries.....
T family → best for low-throughput workloads or instances that sit idle most of the day.
M family → balanced compute/memory for general workloads.
R family → more ram per vCPU, great for memory-intensive apps....
T family → best for low-throughput workloads or instances that sit idle most of the day.
M family → balanced compute/memory for general workloads.
R family → more ram per vCPU, great for memory-intensive apps....
serverlessadvocate.substack.com/p/21-product...
This week, our serverless expert is #AWS Community Builder @ziedbentahar.bsky.social, and our spotlight falls on @danilop.bsky.social 👏
Please share & subscribe to the newsletter!🔔
serverlessadvocate.substack.com/p/21-product...
This week, our serverless expert is #AWS Community Builder @ziedbentahar.bsky.social, and our spotlight falls on @danilop.bsky.social 👏
Please share & subscribe to the newsletter!🔔
You can enable provisioned concurrency to keep execution contexts warm.
You'll pay $0.0000041667 per GB-second for the privilege, used or not - but zero cold starts if provisioned > required contexts.
Your latency-sensitive apps will thank you 🚀
You can enable provisioned concurrency to keep execution contexts warm.
You'll pay $0.0000041667 per GB-second for the privilege, used or not - but zero cold starts if provisioned > required contexts.
Your latency-sensitive apps will thank you 🚀
Focus on rapid iteration, market feedback, and distribution.
3 tips:
- Solve one problem.
- Iterate with real user insights.
- Launch fast, pivot faster.
calvintorra.com/blog/saas/wh...
Focus on rapid iteration, market feedback, and distribution.
3 tips:
- Solve one problem.
- Iterate with real user insights.
- Launch fast, pivot faster.
calvintorra.com/blog/saas/wh...
newsletter.eng-leadership.com/p/from-laid-...
As someone recently laid off and looking for a new role I've also considered tracking applications, but the amount of ghosting makes it feel meh.
newsletter.eng-leadership.com/p/from-laid-...
As someone recently laid off and looking for a new role I've also considered tracking applications, but the amount of ghosting makes it feel meh.