We are constantly running performance tests to ensure that Epsagon will not slow down your application. See more information below.
Epsagon deploys Trace Collector in all regions and AZs. These are extremely fast services which merely collect traces and return response right away. According to our testing, Epsagon's tracing impact is:
- In containerized environments - limited memory and CPU consumption (<1%), with smart batching and flushing.
- In Lambda functions - add between 1 to 4 milliseconds to every call on average.
There is no price impact, as our trace collectors are deployed in every cloud region.
We constantly make sure to limit Epsagon's library size as to not inflate your code. It is verified to be around 100 KB.
Since Epsagon only adds 1 to 4 milliseconds to each call, it's very unlikely that you will get a timeout because of Epsagon. If you did get a timeout, then you would probably get the same timeout regardless of Epsagon (BTW - Epsagon can greatly help you fix that!).
AWS Lambda is billed in whole units of 100 milliseconds, and therefore will almost always not be affected at all by the Epsagon agent. Even if affected, Epsagon only adds a few milliseconds, and the impact on the cost will be negligible.
Epsagon only sends small amounts of data that are embedded in the trace.
Updated 10 months ago