Mark_NL
3rd November 2010, 09:18
I would like to ask for your experience on clustering systems.
I've got 2 identical servers with both 44TB (24x2TB RAID5 on a Qlogic fiber card) and am looking for a possible clustering solution.
The raid will most likely contain large files (re-encoded movies, dvd images, etc.) (yes these are legal).
I want to keep both machines in sync at all times, and want our customers to be able to download files from both servers (we use loadbalancing to spread the downloads). We will upload new movies to only 1 servers and should be replicated over the cluster automatically (or when it's requested).
I've only been testing a little bit with GlusterFS, which looked oke, but haven't really stress tested it (yet). Though i'm wondering if clustering is suitable for these large files. (Ofcourse they will are connected with a gbit eth card and if needed i can aggregate them)
Any of you have experiences with other clustering systems? Or is GFS the way to go?
p.s. I'd like to start with a 44tb cluster and as it grows add more filers to expand the cluster.
I've got 2 identical servers with both 44TB (24x2TB RAID5 on a Qlogic fiber card) and am looking for a possible clustering solution.
The raid will most likely contain large files (re-encoded movies, dvd images, etc.) (yes these are legal).
I want to keep both machines in sync at all times, and want our customers to be able to download files from both servers (we use loadbalancing to spread the downloads). We will upload new movies to only 1 servers and should be replicated over the cluster automatically (or when it's requested).
I've only been testing a little bit with GlusterFS, which looked oke, but haven't really stress tested it (yet). Though i'm wondering if clustering is suitable for these large files. (Ofcourse they will are connected with a gbit eth card and if needed i can aggregate them)
Any of you have experiences with other clustering systems? Or is GFS the way to go?
p.s. I'd like to start with a 44tb cluster and as it grows add more filers to expand the cluster.