Main RAM holds writes.

When the area in main RAM reaches its capacity, things are kicked off.

NVRAM is a paralleled part of the life of a write cycle, but it's not a step that a write passes through.

I see the differences in how its explained..both are right..but the original intent, was to remove NVRAM as the culprit of an IO problem.

Short of the battery being dead, or the memory on the card itself failing..its not in play.


On Thu, Apr 11, 2013 at 9:23 PM, Michael Bergman <michael.bergman@ericsson.com> wrote:
Blackburn, James M (Jim) Jr. CIV USARMY PEO EIS (US) wrote:
Classification: UNCLASSIFIED
Caveats: NONE

If "NVRAM has no place in the IO operation of a filer", why doesn't every
NetApp storage system have the same amount of NVRAM/NVMEM?

It all depends on what the definition of 'has a place' is.
I'd argue that the NVRAM is involved in the I/O operation of a Filer.

Putting copies of data (which sits in RAM) into NVRAM, mucking about with it (to keep it 100% consistent at all times in case of something sinister happening at *any* time!), and making sure it's emptied out etc etc kindof puts it in the I/O path w.r.t. capacity. It's an internal limiting factor inside the ONTAP kernel so...

The size of the NVRAM is effectively limited by the desired cluster pair failover time. The contents have to to be mirrored across the CI link all the time as well.  It's an optimisation problem, like so many other things inside a storage system of the complexity that a FAS box has

In the coming new hi-end FAS models (not before early 2014 is my guess), the NVRAM will be larger than now due to other things mitigating the increasing failover time. (It's a whole new HW platform.)

/M
--
Michael Bergman
Sr Systems Analyst / Storage Architect
Engineering Hub Stockholm                Phone +46 10 7152945
Service Delivery, Engineering EMEA N     SMS/MMS +46 70 5480835
Ericsson                                 Torshamnsg 33, 16480 Sthlm, Sweden
--
This communication is confidential. We only send and receive email on the
basis of the terms set out at www.ericsson.com/email_disclaimer

_______________________________________________
Toasters mailing list
Toasters@teaparty.net
http://www.teaparty.net/mailman/listinfo/toasters



--
---
Gustatus Similis Pullus