When I worked for a National Supercomputing Center, I discovered that they had a distributed filesystem called Andrew Filesystem (AFS) which they used for archival purposes. With every new generation of distributed filesystem I always wondered why the older ones failed.
It wasn't until maybe 10 years ago that I finally got my answer: It turns out that Amdahl's Law kills AFS. There's a total throughput wall that becomes very painfully visible once you move to gigabit networking, and any one client can pretty much saturate the network.
Comments
When I worked for a National Supercomputing Center, I discovered that they had a distributed filesystem called Andrew Filesystem (AFS) which they used for archival purposes. With every new generation of distributed filesystem I always wondered why the older ones failed.
It wasn't until maybe 10 years ago that I finally got my answer: It turns out that Amdahl's Law kills AFS. There's a total throughput wall that becomes very painfully visible once you move to gigabit networking, and any one client can pretty much saturate the network.
That is also what CERN was using in early 2000 before the Grid efforts.