I wont replicate my data with SAN.You carve multiple LUNs and distribute it across storage servers. Setup UCARP or CTDB for HA and you are good to go. Synchronous replication was developed to address the limitations of DAS. If a server goes down in a SAN environment, you can migrate LUNs and IPs to another available node. You can't do that with DAS. Disks go down with a server failure. How ever DAS is cheaper and simple when compared to SAN even with replication turned on. Amazon, Google to Facebook replicate their data 3 times and still claim massive scalability and economics. SAN has its own advantage in terms of reduced power, real estate and cooling. It is backed by top hardware vendors. DAS is going to be faster, easier and cheaper in most cases though. On Mon, Apr 4, 2011 at 1:13 PM, Michael McGlothlin <michaelm at plumbersstock.com> wrote: > I have a 10Gb dedicated network for data storage. Do you think it'd > still be a problem? > > Thanks, > Michael McGlothlin > > > > On Mon, Apr 4, 2011 at 1:47 PM, R.C. <milanraf at gmail.com> wrote: >> I don't think this would be a great idea. Not just for the iSCSI targets but >> for ESXi datastore use. When a GlusterFS brick fails and resumes, replica >> reconstruction (given you use it), will eat almost the whole available >> bandwidth, making ESXi fail writing to virtual disks. >> I had the same problem in my test environments. >> >> JM2C >> >> Raf >> >> >> ----- Original Message ----- From: "Michael McGlothlin" >> <michaelm at plumbersstock.com> >> To: <gluster-users at gluster.org> >> Sent: Monday, April 04, 2011 9:33 PM >> Subject: iSCSI? >> >> >>> Has anyone had any success using gluster as a backend for iSCSI >>> targets? I want to create 1TB target files which will be shared by >>> three iSCSI servers for access by ESXi. I am wondering how gluster >>> handles updates to large files such as that - I assume it's smart >>> enough to update only the changed blocks and not try to copy the >>> entire file around on each update? Also wondering how well it will >>> handle multiple servers accessing the same file at the same time. The >>> same blocks, iSCSI files, should never be being written to from >>> different servers at the same time but the target file would be. Will >>> this work? Do I need to do anything special? >>> >>> >>> Thanks, >>> Michael McGlothlin >>> _______________________________________________ >>> Gluster-users mailing list >>> Gluster-users at gluster.org >>> http://gluster.org/cgi-bin/mailman/listinfo/gluster-users >>> >> >> _______________________________________________ >> Gluster-users mailing list >> Gluster-users at gluster.org >> http://gluster.org/cgi-bin/mailman/listinfo/gluster-users >> > _______________________________________________ > Gluster-users mailing list > Gluster-users at gluster.org > http://gluster.org/cgi-bin/mailman/listinfo/gluster-users > -- Anand Babu Periasamy Blog [http://www.unlocksmith.org] Imagination is more important than knowledge --Albert Einstein
