Using SSD disk for views instead a regular one


What is the affect of using SSD disk instead of regular one when using AWS?



The advantage is increased performance. Each node in the cluster is responsible for maintaining and serving view requests for documents that live on that particular node. As views are scatter gathered off disk for a view query, there can be a performance gain by using faster storage hardware, such as SSD’s from AWS.

Isolating the views from the data files at the hardware level can also improve performance.



Can you say how much it is increasing the performance?
Do you have a calculation that compares it to ESB for example?

I asking it because i want to decide if i should spent more money on such a disk.


For the algorithm we use for sizing internally our results indicate roughly 3-4X better IO performance for provisioned iops than EBS. For a highly specialized CAS SDD based system this can be on the order of 15-20x better performance than EBS. These are just rough guidelines for IO based on sizing information we’ve experienced thus far.