Checking latency in DataCore GUI

DataCore SSY-V brings hundreds of counters for performance monitoring. Some of them are quite easy to understand, some are not. Some show useable data, some not. The problem here is to find these counters you can really see what's going on and if you have a problem or not.

If you think your storage is too slow you probably search for latency statistics. In SSY-V latency statistics can be found several times but which one would you use e.g. as a counterpart to the latency statistics shown by esxtop. Perhaps you don't have ESXi in place and have to find those metrics for Windows VMs. Until today I haven't found a good ressource for checking latencies on a Windows machine.

Luckily we have DataCore and that's all we need. Let's dig a bit deeper....

First you have to go to Live Performance and choose the right counter. In our case this is Category: Virtual logical units, Instances: the actively used path between the host HBA and the DCS server port and Counters can be Average time read/write, Total bytes transferred/sec and Total operations/sec.

Virtual logical units is a virtual instance containing the application servers HBA where data is actively sent and received, the frontend port of the used DCS and the LUN ID the vDisk is served to the host. All these combined create a virtual path between server and storage. All metrics are then shown exclusively for that path.

Finding the correct "mapping" isn't that easy. First you should find out which application server you want to monitor. Then check out which HBA is used for data transmission. This can be done in the DataCore MPIO console. If you have Round Robin as path policy all HBAs and ports are actively used.

Second find out which LUN you want to monitor. This is the easiest thing to do. You can use Windows disk manager for example to get the information.

Finding the active frontend port on the DCS is fair as you can use the preferred host setting and then you probably have only one or two frontend ports that can be used.

Last but not least choose the metrics you want to show up. For example if you choose Average time as counter you will see the average time a storage request needs to be requested and fulfilled by the storage for that particular server and vDisk.

During the first tries I always checked all available pathes between my server and a special vdisk because didn't want to search for the right one. This will fill up your 50 available counters quite fast so it's worth to check a bit deeper which virtual logical unit you really want to monitor and then select only counters for that special mapping.

Another option during troubleshooting is to reduce the number of available pathes to only one so you can be sure you get the right one and path policies will not interfere with your measurement. 

Leave your comments

Post comment as a guest

0
Your comments are subjected to administrator's moderation.
  • No comments found
Powered by Komento
joomla templatesfree joomla templatestemplate joomla
2017  v-strange.de   globbers joomla template