Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

I was asking recently about options for divvying up access to file servers, as we have a NAS solution that gets fairly bogged down when our users (with giant profiles, especially) all log in nearly simultaneously. I ran across Gluster and it looks like it can cluster different physical storage media into a single virtual volume and share it out like a virtual NAS from the client perspective and it support CIFS.

My question is whether something like this would be feasible to use for home and profile directories in an active directory environment.

I was worried about ACL's, primarily, as I didn't think CIFS was fine-grained enough to support NTFS permissions and it didn't look like Gluster exports those permission levels, just the base permissions for basic file sharing.

I got the impression that using Gluster would allow for data to be redundant across multiple servers and would speed up access to the files under heavy load, while allowing us to dynamically boost storage capacity by just adding another server and telling Gluster's master node to add that server. Maybe I'm wrong with my understanding of it though. Anyone else use it or care to share how feasible this is?

share|improve this question
    
Did you end up choosing Gluster? What is your current NAS setup? We are looking at Nexenta but are thinking about testing Gluster... –  tsquillario Sep 16 '11 at 13:32
    
Not using Gluster; ended up getting another server with higher-speed drives to help with home directory server. Boss didn't want to purchase additional hardware for a file serving cluster. –  Bart Silverstrim Sep 16 '11 at 13:39
add comment

1 Answer

up vote 1 down vote accepted

Theorically, GlusterFS is an answer to your need.

Using GlusterFS, you can easily create RAID0 (type cluster/distribute) and RAID1-like volumes (type cluster/replicate), distributed across many machines.

GlusterFS architecture lets you stack translators in a way you can create 2 distributed volumes, replicate them, and then, access your distributed/replicated data through an unique mount point.

However, there are some feedbacks of bugs appearing when users stack such translators (see GlusterFS mailing lists). That's why I don't trust GlusterFS enough to setup RAID10-like volumes. (Since I don't tested this setup enough, it's only a belief)

Of course, simple RAID0-like and RAID1-like volumes seem production-ready.

share|improve this answer
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.