EC2 - VolumeQueueLength

Question

Hi guys,   
  
I have an Amazon EC2 instance (micro) that generates ZIP files and uploads them to S3. For the past few years I haven't really had any issues but recently the instance has occasionally been crashing during busier periods when more ZIPs are being generated.    
  
I think I will resize the instance to medium but wondering if the volume attached to the instance will need to be increased as well? The current volume is a General Purpose SSD 30GB 100 / 3000.Queue Length Graph during busy period here: https://ibb.co/0FF7VX3  
  
Thanks  
  
Edited by: ajh18 on Sep 16, 2019 8:48 PM  
  
Edited by: ajh18 on Sep 16, 2019 11:29 PM

Answer

Hi,   
I think the answer is "it depends". The VolumeQueueLength during the peak times is definitely showing a bottleneck. However, because you stated that your "Zips" are crashing, the crashes could possibly be due to memory shortage, which may result in "page file swapping", which may result in overloading your disk I/O to swap in/out chunks of process memory.  So, it is possible that by moving to a t2.medium, you will now have adequate RAM so that swap will be 0 or minimal, which may result in no Disk I/O bottlenecks.  
To keep costs minimal, I would take the following approach.

```
1. Increase the instance size from t2.micro to t2.medium
2. Monitor for performance, VolumeQueueLength, and reliability
3. If you are still noticing performance issues, and the VolumeQueueLength continues to show a bottleneck during the peak times, then increase the Volume size (which will increase the IOPs).
```

Hope this helps,  
-randy

Answer

Thanks Randy! I'll resize to medium and continue to monitor.

EC2 - VolumeQueueLength

Relevant content