By erickustarz on Mar 04, 2005
So now that Connectathon is over, what was the biggest highlight?
I would have to say having the first successful demonstration of pNFS (courtesy of Sun and Netapp). The three amigos from Sun (Sam Faulkner, Lisa Week, and Alok ) ran a Solaris client against Netapp's server (Garth Goodson). Garth later on gave a talk about it. The basic premise is for very large files, having only one server can cause us to run into bandwith limitations. Striping a very large file across several (in the presentations there are ALWAYS three) data servers, and then having that "layout" information passed back to the client via the meta-data server, and then all reads/writes are sent from the client to the data servers can solve the bandwith problems (and lets you stay in vogue with horizontal scaling).
The internet draft for pNFS can be found here.
This demonstration used files/NFSv4 as the backend protocol to move the actual data (as opposed to just talking to the meta-data server). But you can easily see how the block people are excited to use their own solutions. And can also see how the cluster people are very interested in an open standard.
Ok, so its parallel... why should i care? To quote Goodson:"
This draft considers the problem of limited bandwidth to NFS servers. The bandwidth limitation exists because an NFS server has limited network, CPU, memory and disk I/O resources. Yet, access to any one file system through the NFSv4 protocol requires that a single server be accessed. While NFSv4 allows file system migration, it does not provide a mechanism that supports multiple servers simultaneously exporting a single writable file system.
This problem has become aggravated in recent years with the advent of very cheap and easily expanded clusters of application servers that are also NFS clients. The aggregate bandwidth demands of such clustered clients, typically working on a shared data set preferentially stored in a single file system, can increase much more quickly than the bandwidth of any server. The proposed solution is to provide for the parallelization of file services, by enhancing NFSv4 in a minor version."
So now it works, excellent. Sam, Lisa, Alok... when are we going to get perf numbers? heh
Lastly, i'd like to point out that this shows the beauty behind NFSv4. This is exactly what v4 was designed for, openness and enhancements (via minor versions) where everyone can play. We can build upon the base of the protocol to solve interesting problems instead of redoing the (un-interesting) work every five or ten years.