4

I have an ALB that is performing a health check every 30 seconds on a Target Group (Lambda endpoint in this case). Despite this, any time I query my API for the first time in several minutes or longer it goes through a cold start. What could be the reason? Do ALB health check initiations of Lambda functions intentionally bring up a separate isolated instance purely for health checking, and are not the same instances used by a general query to the URL? CloudWatch seems to confirm my suspicions because ALB checks are all logged to separate log streams within the log group of the lambda. Thanks in advance for your help.

anonymous developer
  • 398
  • 3
  • 17
  • Have you got the reason yet ?, I'm writing slack bot and slack has a restrict with 3000ms response time. I have to move my slack bot to Google App Engine. – vanduc1102 Mar 03 '20 at 04:40
  • 2
    Yes, the reason is that AWS uses a specialized algorithm that keeps lambda's warm and determines the speed of the response time by the number of requests being sent in a given interval as well as the number of concurrent requests. If you wish to keep your lambda's warm you must send concurrent requests at repeated small intervals (the number of concurrent requests and the size of this interval depends on how fast you wish the response time to be, and there is no public documentation on this except for benchmark tests run by 3rd parties). – anonymous developer Mar 05 '20 at 18:26

0 Answers0