June 10, 2015 at 8:12 am
A customer has reported speed issues in our application Since I don’t understand the below statistics that are mentioned below such as the reads, writes, waits, and virtual file latencies,
can anyone please explain what those are and if any of those might be anything that would worry you?
below is the information about SQL Server of client:
CrystalDiskMark drive information on the server (where we did the timings) is as follows:
-----------------------------------------------------------------------
CrystalDiskMark 3.0.3 x64 (C) 2007-2013 hiyohiyo
-----------------------------------------------------------------------
* MB/s = 1,000,000 byte/s [SATA/300 = 300,000,000 byte/s]
Sequential Read : 559.763 MB/s
Sequential Write : 177.981 MB/s
Random Read 512KB : 75.734 MB/s
Random Write 512KB : 108.795 MB/s
Random Read 4KB (QD=1) : 1.079 MB/s [ 263.5 IOPS]
Random Write 4KB (QD=1) : 5.830 MB/s [ 1423.4 IOPS]
Random Read 4KB (QD=32) : 8.117 MB/s [ 1981.8 IOPS]
Random Write 4KB (QD=32) : 8.451 MB/s [ 2063.1 IOPS]
Test : 1000 MB [C: 67.7% (67.7/100.0 GB)] (x1)
Date : 2015/06/02 15:18:44
OS : Windows Server 2008 Server Standard (full installation) SP2 [6.0 Build 6002] (x64)
According to a script I have the top waits according to SQL are of type (in decending order):
CXPACKET @ 4592.31
PAGEIOLATCH_SH @ 1716.63
ASYNC_NETWORK_IO @ 1225.21
According to another of the scripts I have the virtual file latency in (descending order):
master.mdf
msdbdata.mdf
model.mdf
modellog.ldf
msdblog.ldf
tempdb.mdf
templog.ldf
mastlog.ldf
June 12, 2015 at 11:17 am
Given that CXPACKET waits was larger than the other waits, thought I'd go looking for good info, and found some from a really good source:
http://www.brentozar.com/archive/2013/08/what-is-the-cxpacket-wait-type-and-how-do-you-reduce-it/[/url]
Read it, and toward the end, there's a link there for more on how to tune. Enjoy!
Steve (aka sgmunson) 🙂 🙂 🙂
Rent Servers for Income (picks and shovels strategy)
Viewing 2 posts - 1 through 1 (of 1 total)
You must be logged in to reply to this topic. Login to reply