Dvir,
Thanks for replying despite the vagueness. Here is a more detailed situation I am facing.
My application is deployed on a VM and has a functionality to read close to 10K key/values from Redis (Either at once or in buckets using pipelines or individually). Each key/value is around 100 KB. Now because of this, the data transferred on the network is going up. While Redis responds fast, the high rate of network transfer is causing issues of a different kind.
The VM where the application has been deployed is a busy VM with 70% CPU on an average.
We were considering bringing in the Redis on the same VM where the application is installed so that the network data transfer can be avoided. We could have used local cache in the application, but presently it is not an option.
So my question was - Since Redis is a single threaded cache (almost!) and is very CPU sensitive, was wondering if it is a good idea to move the Redis into the VMs that see a lot of activity.
I can probably benchmark it to figure out, but wanted to get some opinions from the community too.
Thanks
Kashyap