We have a VPC setup on Amazon S3 where we have various application delivery web servers that communicate with a cluster of NCache in-memory caching servers, and a database server.
Trying to determine the bottleneck and how to properly scale; would it be possible and a good idea to create multiple NICs per instance and segment them so that a (or load balanced cluster) of NIC(s) on the web servers are load balanced, and then another dedicated NIC for traffic to the cache servers network, and another for the database.
Right now from monitoring on AWS, our cache servers are seeing around 600Mbps in and 1.5Gbps out, with very minimal load.
Aucun commentaire:
Enregistrer un commentaire