One thing to watch for with Filers and clustering is the vram size,
especially under high write loads. The clustering effectively halves
the vram each head can use. That means more cycles going to clear out
the cache to disk which is the filers slowest operation.
We ended up unclustering our heads and doubling the vram. It had a
dramatic impact on our write speed which in turn improved the read
latency problems we were encountering.
That being said, NetApp's support and trouble shooting assistance has
been exceptional.
-- John
Rob Groome wrote:
[posted to rush.general]
I would tend to agree with what Dylan said here. I looked at the EMC
product and was not impressed with the support that I got pre-sale - so
I was not really that interested to find out what their post-sale
support would be.
I did much legwork to get as much information about both NetApp and EMC,
and in everything I found, the NetApp stuff just kicked ass over the
EMC, especially in support.
We purchased a clustered 3020 that is currently sitting in my office
waiting to be setup this week. :)
Rob
--
John Pierson 503-595-4070
R&D Lead Enginerd 503-464-9555 FAX
Laika Entertainment jpierson@(email surpressed)
|