Ask Your Question

openstack swift deployment model for read intensive environment

asked 2013-11-07 08:18:16 -0500

fga gravatar image

updated 2014-01-24 00:39:42 -0500


We are currently trying to make capacity and resource planing for a read intensive environment in which 5000 concurrent connection need to be established. While considering number of servers required for proxy-server and storage-server(account,container and object in the node), we are thinking of making a model with proxy-server count > storage-server count due to possible high load in proxy-server during queries for object and download being streamed from object server to user through proxy-server and in front of proxy-server we will have redundant loadbalancers.

Is this the right approach for having more proxy-servers than storage-servers in deployment model, considering read intensive environment?


Update: For more details our use case clients are downloading the same 250MB archieve file.

edit retag flag offensive close merge delete


Could you have more detail about your case? * How's the average size of each object * The *Read intensive* means read same object or different objects concurrently ? The bottle-neck will appear on proxy either storage drives counts. You need a to have base environment to measure the performance of the kind of deployment. And them scale out to cross the bottle-neck.

Hugo Kuo gravatar imageHugo Kuo ( 2013-11-13 22:05:04 -0500 )edit

1 answer

Sort by ยป oldest newest most voted

answered 2013-11-12 22:38:02 -0500

koolhead17 gravatar image

Yes more proxy server will be helpful.

edit flag offensive delete link more

Get to know Ask OpenStack

Resources for moderators

Question Tools

1 follower


Asked: 2013-11-07 08:18:16 -0500

Seen: 216 times

Last updated: Jan 24 '14