> It will be interesting to see if a different picture
> of the Chameleon product emerges.
It is pretty simple: EMC continues to introduce performance
benchmarks that don't include RAID (parity) protection. Both
their high-end "Celerra" and mid-range "IP4700" have been
benchmarked with RAID turned off. Why?
In addition to hiding performance problems by turning RAID
off, EMC also does the following benchmarking hijinks:
1. With Celerra, adds tons of memory to the attached Symmetrix
to fudge their numbers up. Note the following from a SFS entry:
"Special Config Notes - 16GB memory per Symmetrix"
See: specbench.org
2. With Celerra and IP4700, add extra hardware (RAID controller
or disks) to help fudge numbers. Note that latest IP4700 system
was configured with 100 disks, the latest NTAP F840 numbers were
generated with 18 disks. Why not compare with similar configuration?
3. New with latest IP4700 numbers: use 4-processor Sun systems with
4 gigabytes of memory each to get clients responding as quickly as
possible. This is a new twist, I don't think I have seen this one
previously.
If I were EMC, the analyst-appointed "Storage Gorilla", and I wanted to
introduce a "NTAP-killer", I think I would configure a system
similar to NTAP and show head-to-head that it was a better system.
EMC chose not to do this because this product is not a "NTAP-killer",
nor a "Sun-killer" and probably not even an "Auspex-killer".
If there are any analysts reading this note, please print out its
context and ask EMC to comment. They are flat-out-lying about the
capabilities of this product, never mind the Celerra. |
