05-09-2009 05:14 AM
My NB environment is NB 6.0 MP6 on a windows 2003 server standard edition.
The server has 4 Emulex LightPulse Fiber adapters ( 2 on each HBA ) (Unique ID's 0D70, 0D71 & 37C4,37C5) connecting to 2 EMC McData Fabric switches.
The SAN consists of an EMC Clariion CX3 raid array and an SL500 Robotic tape library.
The SL500 has 4 tape drives, each drive is connected to each switch as well as the robot (Unique ID's Robot D1DD, Tape drives D1EA, D1E7 & D1ED,D1F0)
The Clariion has 2 SP's with 2 adapters connecting to each switch (Unique ID;s 6041E018EE,6841E018EE & 6841E018EE,6941E0188)
The problem I am having is a severe slow down to the point where the server reaches 100% cpu utilization. It seems to happen on the bigger clients where the data is over 300 gigs. The longer it takes the worse it gets. Staging jobs (DSSU) seem to make it worse.
I was hoping some one could lend me some advice on proper zonning of the fabric switch. Currently I have 1 zone for the Clariion and 2 adapters for the master server and a zone for the master server and the tape drives/robot.
example of my current zones
zone 1: master server -1 adapter from each HBA
clariion - 1 adapter from each SP
master - 0D70
master - 37C4
clariion - 6141E018EE
Clariion - 6841E018EE
zone 2: master server - 0D70
robot - D1DD
tape drive - D1EA
tape drive - D1E7
I am wondering if i need to isolate the master server adapters to only 1 zone. For example, remove 07D0 from the Clariion zone.
If any body has any advice on this, I would certainly appreciate the help.
thank you
dave
05-09-2009 06:25 AM
05-09-2009 06:33 AM
05-09-2009 06:38 AM
05-11-2009 07:34 AM