/plushcap/analysis/datadog/datadog-wild-fixing-slow-writes-aws-elastic-block-store

Datadog in the wild: Fixing slow writes on AWS Elastic Block Store

What's this blog post about?

Konotor, a mobile-first user engagement platform, faced challenges in scaling their PostgreSQL database due to slow writes on AWS Elastic Block Store (EBS). They discovered that EBS was throttling their writes as they had failed to correctly calibrate how many IOPS the disks would consume. Using Datadog, Konotor investigated and identified the issue, leading them to explore various solutions such as striping multiple disks, using Provisioned IOPS (SSD), ephemeral storage, moving frequently used tables/indexes to another disk, and reducing I/O writes. They also emphasized the importance of monitoring and alerting for I/O transactions, CPU time spent on I/O waits, and disk latencies. Datadog's ability to correlate data, gather metrics from various sources, and its lightweight nature made it an invaluable tool for Konotor during this process.

Company
Datadog

Date published
March 3, 2015

Author(s)
Deepak Balasubramanyam

Word count
1119

Language
English

Hacker News points
None found.


By Matt Makai. 2021-2024.