[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [Xen-users] Multi-server setup, storage question
Hi,To make life easier, this e-mail is divided in three sections (background/specs, question, goal). First some information that might be useful to the question further down. I have two identical servers as per this config: - 8 cores each - 64 GB ram each - HW-RAID backed storage (~695 GB/SAS 15k RAID6) - Quad Gigabit NIC - Linux 18.1 (currently) - Xen 4.4.1 from openstack's gplhost repos (currently) - Debian Wheezy with Backports (currently)I am looking to run all the XEN guests from the local hw-raid storage and use LVM for the guest disks. I am using local storage as opposed to an iSCSI SAN since performance / latency is very important in this project. Now for the question:Is it a good idea to use a distributed file-system to make "one big slab" of LVM storage and then set up guests disks on that? Or should I use DRBD to mirror the whole storage between the servers?Ideally I would want to be able to use both machines to run guests at the same time, but I am quite happy to have one server on "hot standby" too. But I am unsure about the performance impact a distributed file-system would have? And which one to choose if that was the case? The goal: To have full hardware fail-over.Let's say server 1 has a catastrophic failure, then I want to be able to retain all the data and start the guests on server 2. DRBD would solve this, since it could mirror the whole "storage surface" between server A and server B. But if I had a distributed filesystem, then ideally I could run guests on both server A and B at the same time. And if server A or B died, I would still have all the data and be able to run the guests on the surviving server (since I will not over-provision storage over the ~695GB available). -- Henric Blomgren _______________________________________________ Xen-users mailing list Xen-users@xxxxxxxxxxxxx http://lists.xen.org/xen-users
|
Lists.xenproject.org is hosted with RackSpace, monitoring our |