[stgt] cluster support (the next major feature)
fujita.tomonori at lab.ntt.co.jp
Wed Jun 9 04:42:50 CEST 2010
I've been thinking about adding the feature to build a stgt cluster. I
want to build a storage system like Lefthand  or PeerStorage 
- The cluster consists of several nodes. Each node runs stgt.
- One of the nodes has a virtual cluster IP address (if the node
crashes, another node takes over the address).
- Initiators always connect to the virtual cluster IP, then are
redirected to one of the nodes (thanks to iSCSI).
- stgt stores the data in several nodes.
- No single point of failure.
- you can scale the capacity by adding a new node to the cluster.
I know that to achieve the above, stgt might use "cloud
infrastructure" such as hail, zookeeper and hdfs, etc. However, I
need the solution now. It takes long time since we have such code in
various distributions (especially, enterprise distributions).
I plan to use corosync for node management and Berkeley DB
(replication feature) for metadata. Both is mature and shipped with
major distributions. They don't scale with hundreds nodes though.
I did the similar things (and more) in Sheepdog project. I want to
do less ambitiously and more quickly with the existing protocol
Opinions? The existing options?
To unsubscribe from this list: send the line "unsubscribe stgt" in
the body of a message to majordomo at vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
More information about the stgt