Unsolved
This post is more than 5 years old
103 Posts
0
1516
High LUN Utilization and Response Time
Hopefully you guys wont get tired of me asking to many performance questions.
Again, I noticed a SQL LUN (houses the Index) and noticed that at one point:
:i had over 350ms of RT and the LUN was at 100% utilization
:The Total Throughput during this time was pretty low, so I know that I have enough spindles to accommodate up to my MAX IOPs
:The LUN was doing alot of reads
:The SP wasnt doing any flushing (currently between the default watermark setting)
:disk crossings (%) was between 100-130 level during this time
:Queue Length was between 7-10 during this time
So what gives? What could be my bottleneck? could it be my spindle count?
joej1_3ddf8c
34 Posts
1
February 18th, 2010 09:00
One thing to keep in mind is response time is a measured and is inversely proportional to LUN throughput and directly proportional to the q-length at that instant. Low throughput will aritificially report high response times. You should verify the response time as seen from the host/application.
Cheers,
-joji
kelleg
4.5K Posts
0
February 19th, 2010 08:00
If you look in the Navisphere On-Line Help - Analyzing Storage-System Performance using Analyzer/Troubleshooting/Troubleshooting Performance Metrics - find the section called "Lun Response Time" - that explains what you are seeing.
In addition, there is a lot of very good information in the on-line help about using Analyzer and how to interpret the different metrics. EMC also offers a Performance Class.
glen
h1pan1
18 Posts
0
July 15th, 2010 06:00
Cool Joji:>)
Henry