Unsolved
This post is more than 5 years old
1 Rookie
•
79 Posts
0
8709
July 25th, 2017 21:00
How to check the performance of Equallogic?
Hello friends,
I have an Equallogic PS4100X which is connected to a two switches. These switches are connected with link aggregation (3 x 1Gb cables).
I am using RAID50 with 24 disks.
On these same switches, I have VLANs to track network traffic. Isolating including the ISCSI VLAN and the VLAN from the server network.
My entire VMware hosts connection network is 1Gb.
I have LUNs for groups of virtual, neighboring servers to avoid allocating all VMs on a single LUN LARGE.
Some virtual servers use provisioned disks and some fixed disks.
My environment is a VMware 5.5.
I have noticed a slow startup of virtual servers and even some applications.
I would like to know how I see how the storage performance is. I'm not sure if I check into VMware or if it's on Equallogic.
I think I'm starting to have a performance problem with Equallogic, but I'd like to see that.
Thank you.



Sandro Silva Alves
1 Rookie
•
79 Posts
0
July 25th, 2017 23:00
Sorry for the incorrect translation.
My VMs are separated into small LUNs.
What I meant is that I DO NOT use large LUNs. I try to avoid ...
LUN1 (1TB) = VM1 (Disk1 C: \ and Disk 2 D: \), VM2 (Disk1 C: \), VM3 (Disk1 C: \ and Disk 2 E: \)
LUN2 (500GB) = VM4 (Disk1 C: \ and Disk 2 D: \), VM5 (Disk1 C: \), VM6 (Disk1 C: \ and Disk 2 E: \)
LUN3 (500GB) = VM7 (Disk1 C: \ and Disk 2 D: \), VM8 (Disk1 C: \), VM9 (Disk1 C: \)
LUN 4 (500GB) = Physical Server (Database - DB)
LUN 5 (100GB) = Physical Server (Database - DB)
LUN 6 (20GB) = Physical Server (Database - Log)
The scenario is something in that format.
I do not know SANHQ, but I looked at Blogs now and saw that it shows a lot of information. Is there any reference I can look at to know if I'm overloaded?
I already used VMware ESXTOP and it helped me a lot to find some problems. But now, I think my environment has grown and I need to show indicators that the physical environment is overloaded.
I want to start with the storage to see if the transfer is actually as expected. For, as we use ISCSi, the performance is directly linked to the input and output of data from the virtual disks.
Thank you.
Origin3k
4 Operator
•
2.3K Posts
0
July 25th, 2017 23:00
I thought he already have multible Volumes = ESXi Datastores instead of a single large one.
Install SANHQ to see Latency/QueueDepth informations. There is a feature called "Live View" in SANHQ which shows some sort of "real time" statistics for a given volume.
On the ESX use "esxtop" to check latency of the device/VMs in realtime. vCenter/vRO statistics may help also.
Regards,
Joerg
Sandro Silva Alves
1 Rookie
•
79 Posts
1
July 28th, 2017 16:00
Hello,
I did an analysis and really these options are different from the good practices.
I will apply these settings to:
- LoginTimout to 60 of the iSCSI adapter;
- NOOPTimout to 30 of the iSCSI adapter;
- Delayed ACK to clear the iSCSI adapter;
A doubt:
Do these settings have any relation to the firmware version of my Equallogic? Can I change these settings with the version of my firmware that is 8.0.6?
Thank you.
Sandro Silva Alves
1 Rookie
•
79 Posts
0
July 28th, 2017 18:00
Hello,
got it. So, I'll do the settings today.
My VMs are configured like this:
- SCSI controller 0: LSI Logic SAS
- Hard Disk: Virtual Device Node: SCSI (0: 0) Hard disk 1
I do not work with RDM disk.
When I need to add another disk, I usually use the same one (SCSI controller 0).
- SCSI controller 0: LSI Logic SAS
- Hard Disk: Virtual Device Node: SCSI (0: 0) Hard disk 1
- Hard Disk: Virtual Device Node: SCSI (0: 1) Hard disk 2
All in all I have 8 VMs per physical host at most.
Do I need to make changes to the VMs too, to gain more performance?
I looked at latency using SANHQ and it shows:
- Lantency: Between 10ms and 20ms, with peaks of 30ms
- Avg Queue Depth: Between 3 and 6
I also have a physical server, connected in Equallogic with iSCSI and two more hyperv core host with 5 VMs.
Thank you.