Looking for IRIX or Solaris expertise? Visit my UNIX Consultancy website.
As systems scale and data volumes grow, access to, and manipulation of that data becomes a serious bottleneck.
I’ve found that many people are unaware of CXFS, or it’s benefits, and in general there’s a fair amount of confusion over the difference between concurrent access to a SAN filesystem versus shared cluster filesystems, as provided by something like Clustered VxFS.
So, I’ve collected together some documents and information which will give a good grounding on CXFS, and hopefully help show what an impressive feat it is.
DMF - Data Migration Facility - is a valuable tool that can live off the back of CXFS. You might be familiar with the concepts of HSM - Hierarchical Storage Management. DMF is pretty much the same thing.
- SGI’s CXFS page
- CXFS Administration Guide from Techpubs
- Local copies of:
- Los Alamos National Laboratory CXFS briefing
R. Kent Koeniger’s April 1999 briefing - a very good overview - Backing up and Restoring multi-terabyte datasets A white paper where SGI demostrate 10TB per hour backup throughput
- NCAR Visualisation and Enabling Technologies
A report detailing CXFS performance - SGI Mobile Innovation Centre at University of Waterloo
Notes from SGI’s MIC visit to UWaterloo in September 2003 - details of CXFS and DMF - Scaleability and Performance in Modern Filesystems
An SGI paper comparing UFS, XFS, VxFS and NTFS - a good overview of XFS and how it works, and how that compares to other common filesystems