MySQL Forums
Forum List  »  Performance

Hints on what to check for bad performance
Posted by: Ulf Zimmermann
Date: September 21, 2015 08:47PM

We have a current installation where we have a master and a number of slaves, which are used for read-only queries. The current hardware older Dell PowerEdge 2850 and 2950 with internal PCI-E Virident SSD cards.

We also have a similar setup for DR, which has HP DL360p Gen8, i.e. a more recent CPUs and memory configuration.

The company is trying to move this environment now to a VMware environment, where the storage is Pure All-Flash-Array. The ESXi servers are currently HP BL680 with X7560 CPUs.

From our application we tried a number of queries, which take about 2 minutes on the current older platform. On the new virtualized environment this takes 5-10 minutes. We are not observing any IO wait, but also no high throughput.

I have run now sql-bench on the DR setup and the virtualized environment. Removing LVM, VMDK layer and putting MySQL data on a file system is slightly faster but not by much. An iozone or dd we are getting well over 100MByte/sec throughput, while MySQL never gets above 25MByte/sec.

The OS is RedHat 6.6 64-bit.

Does anyone have any suggestion what we could look at to find the performance bottleneck?

Options: ReplyQuote


Subject
Views
Written By
Posted
Hints on what to check for bad performance
1393
September 21, 2015 08:47PM


Sorry, you can't reply to this topic. It has been closed.

Content reproduced on this site is the property of the respective copyright holders. It is not reviewed in advance by Oracle and does not necessarily represent the opinion of Oracle or any other party.