San Problem Determination - IBM Storwize V7000 Unified Problem Determination Manual

SAN problem determination

About this task
SAN failures might cause Storwize V7000 Unified volumes to be inaccessible to
host systems. Failures can be caused by SAN configuration changes or by
hardware failures in SAN components.
The following list identifies some of the hardware that might cause failures:
v Power, fan, or cooling switch
v Application-specific integrated circuits
v Installed small form-factor pluggable (SFP) transceiver
v Fiber-optic cables
Perform the following steps if you were sent here from the error codes:
Procedure
1. Verify that the power is turned on to all switches and storage controllers that
2. Verify that the Fibre Channel cables that connect the systems to the switches
3. If you have a SAN management tool, use that tool to view the SAN topology
iSCSI performance analysis and tuning
This procedure provides a solution for iSCSI host performance problems while
connected to Storwize V7000 Unified system, and its connectivity to the network
switch.
About this task
Some of the attributes and host parameters that might affect iSCSI performance:
v Transmission Control Protocol (TCP) Delayed ACK
v Ethernet jumbo frame
v Network bottleneck or over subscription
v iSCSI session login balance
Procedure
1. Disable the TCP delayed acknowledgment feature.
242
Storwize V7000 Unified: Problem Determination Guide 2073-720
the Storwize V7000 Unified system uses, and that they are not reporting any
hardware failures. If problems are found, resolve those problems before
proceeding further.
are securely connected.
and isolate the failing component.
To disable this feature, please refer to OS/platform documentation.
v VMWare: http://kb.vmware.com/selfservice/microsites/
search.do?language=en_US&cmd=displayKC&externalId=1002598(TCP)
v Windows: http://support.microsoft.com/kb/823764
The primary signature of this issue: read performance is significantly lower
than write performance. Transmission Control Protocol (TCP) delayed
acknowledgment is a technique used by some implementations of the TCP in
an effort to improve network performance. However, in this scenario where the
number of outstanding I/O is 1, the technique can significantly reduce I/O
performance.