Administrator Guide

• An action takes too long to complete.
• An action did not nish on time.
Follow these guidelines to analyze the performance problem:
• Are users getting the response time they expect? If not, identify which area might be causing the problem:
– Operating system problem, as it interacts with storage
– Network problem
– Application being run or accessed
– Storage environment
• Use the 80/20 rule. By focusing on 20 percent of the most likely causes of a performance issue, you will solve 80 percent of the
problems.
• Keep the host perspective in mind when managing the arrays. In particular:
– Make sure time is synchronized on all monitoring areas (host, array).
– Never assume the cause of a problem. Other issues might be causing skewed data.
General best practices for solving performance problems include:
• Fix hardware problems immediately, even if the array that is down is the redundant array. Replace failed disks or failed control
modules.
• Always validate the data you are analyzing to ensure it is accurate. For example, you might need to inspect the physical array to
see if it is powered on or check if your switches are connected properly to handle MPIO.
• Consider the size of your installation when analyzing your storage data. For example, an enterprise-level installation with a large
volume of users and data will experience a greater impact from a small degradation of IOPS than a small company with few
users.
• When setting up email notication, determine what types of information are most useful. Streamline the information you receive
as much as possible. If email notications are set too broadly, an actual problem might be obscured by too much information.
• When using SAN Headquarters to monitor your groups, remember that the historical data collected degrades over time. Use
more current data for your analysis.
• Look at event and audit logs for other issues.
Common Hardware Issues
Identifying hardware performance issues can eliminate additional eort elsewhere. Hardware failures can also be a source of
performance problems. In addition, the combination of hardware and rmware can aect performance, as can various disk types
with dierent performance characteristics.
The basic steps for solving any IT problem also apply to the SAN. Table 61. Hardware Issues Aecting SAN Performance lists some
common problems that you should watch for and correct immediately.
Table 61. Hardware Issues Aecting SAN Performance
Damaged
Hardware
Typical Symptom Detected By Possible Corrective Actions
Server NIC Malformed
packets
Monitor errors at switch Update NIC drivers
Replace NIC
Bad cable
Wrong class of
cables
Visible damage
Malformed
packets
Visual inspection
Monitor errors at switch
Replace cable
Defective switch Spontaneous
restarts
Random lockup
Monitor switch with appropriate network Update switch rmware
Replace switch
342
About Monitoring