The Analyse section provides 3 views into your storage data. You can use real time and diagnostic views to debug and find causes for performance problems. You can then use historical to show long term trends and plan for future needs.
- Real Time – Collects Data between 2 and 5 seconds for a limited set of metrics at an array level and is available for 1 hour.
- Diagnostic – Collects data every 5 minutes for Root cause analysis and is available for the previous 7 days.
- Historical – Collects data every 15 minutes and is available for the previous year.
Real Time is focused on an array level view where your focus is on FE/BE Directors and RDF Directors, you can also look at external directors if desired.
As highlighted in red you have the ability to capture a trace if required and display it in the charts section so you have a visual picture of your performance. Bear in mind this is displaying metrics at between 2-5 seconds so it offers you an insight into how the box is behaving at present in a really granular format. To capture the trace you merely select the individual director or all of the directors and then select capture trace.
To display this trace in a chart you merely change views to the charting section and select real time and then trace and you have the option to select 9 KPI’s related to directors so you can focus on metrics like BE IO’s per sec and % Cache WP.
Diagnostic provides high-level data about all the storage systems. The specific information depends on the models of the selected storage arrays.
Highlighted in red to the left we have the category of objects that we can select and drill down on. The list is ranked in terms of how busy the SG based across the values displayed. To the right you have the ability to change the time frame from 4 out to 24 hours. Additionally with this list having over 500 objects you have the ability to filter on it if you require.
I wanted to go over some of the potential use cases also here which will aid you in your troubleshooting. Typically when a storage administrator begins a performance troubleshooting task their starting point from an application perspective is the host.
To do this quickly we can leverage the Analyze section to select hosts and our scope is the last 12 hours, in this particular example the application owner is running on host irco2sd15 and although the host IO’s/Sec and host MBs/Sec are high the host is meeting its response time as we are green in the 3rd column.
A very useful section here is the events and alerts piece if selected it gives you a view into what events and alerts occurred in the last 24 hours.
Here we have the ability to go back to the last 24 hours and also customize the view so if we have a large number of configuration changes we can select a certain timeframe such as 2:00 to 2:30 last night. For this example I have received a call that someone has lost access to their host CA and so I can look back to the changes and I notice that someone ran a delete operation on the port group CA which might explain the loss of service.
Finally we have the Historical view which provides high-level data about all storage systems. This allows you to take a more long term view of how certain elements in your storage array have been performing over a longer time frame going back to 12 months. I just wanted to show you 1 potential use case here for a host.
You can see by the breadcrumb selection that I have drilled to focus on the initiator ports 1D and 3D as while investigating a performance problem with a host the application owner mentioned that performance has been slow for some time. Focusing on the response time for these FA’s we can see that over the course of the last 12 months it has been within expected parameters.
I hope you found this helpful, if you have any questions please let me know.