Sign in to follow this  
Brian

Coraid ZX-Series ZFS Scale-Out SAN Released Discussion

Recommended Posts

So, using raidz2 across nodes via ATA-over-Ethernet is kind of cool. It does let them use a RAIN setup rather than RAID.

But I'm not sure what the advantage of that is. The ZX head unit is still

1. a single point of failure

2. a bottleneck, as all data to the clients must pass through it

I guess if you need a huge volume of storage but aren't worried about HA or throughput, then this would work . . .

Share this post


Link to post
Share on other sites

The role of the ZX is to mount (via AoE) the SRX drives (as JBOD) into ZFS vdevs and pool and serve that up as NFS/CIFS.

1. There are two ZX heads at the top of each rack, so they can run in HA mode. If one fails, the requests are routed to the second. At the back end keep in mind that all the disk traffic is L2, so mulipathing is free. You can run two ZX as active/active and a third passive if you like too.

2. The ZX is a gateway, but unlikely to be a bottleneck is most use cases.

3. Read cache is stored on the ZX units (MLC) and ZIL is SLC on the SRXs. Though I've heard tell of a Zeus RAM for the ZIL as well, which I imagine would live on the ZX heads.

That's the beauty of ZFS and AoE, all the flexibility you get in configuring the system.

There are white papers and case studies and such on their website that explain all this.

John.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this