This post is not about Arduino, microcontrollers, Raspberry Pi, FreeBSD and so on.
But this post is about DIY – Do It Yourself.
I am systems engineer by my primary job. And most times I work with EMC storage systems. EMC provides a lot of tools for their products. But we need something different sometimes.
When I need something different I create what I need.
In this post I will tell about NarReport – my tool for analyzing performance data from EMC CLARiiON/VNX.
EMC tools for analyzing performance data
There are a lot of tools for analyzing performance data from CLARiiON/VNX from EMC:
- Analyzer in the Unisphere
Show performance metrics in real-time or from one archive. You select components and metrics and see the plot. Just plots, no analyzing.
- Analyzer Helper
Old tool, rare used now. Needs a lot of computing resources. Provide PPT with plots. This tool was replaced by Mitrend.
Web service for analyzing performance data from EMC and non-EMC products. You upload data and after some time you get a result – presentation and spreadsheets. Last time presentations become small and doesn’t provide a complex review of the performance data.
- NarBlaster (available for USPEED Guru)
Simple GUI for view data from one data file.
What is wrong with all this tools?
In general: not convenient to use and derive not enough information.
Many of them works only with one data file. We can “merge” some data files into one, but this operation averages the data in the output file.
Extracting performance data
All we needs we can extract from performance data files. Even without using a hidden password we can extract metrics, configuration and relations between components.
Naviseccli – a standard EMC CLI for CLARiiON/VNX extracts all what we need.
Metrics from any numbers of data files we can extract to one CSV file.
Configuration and Relations we can extract into text or XML files.
After that we can analyze this data by any any convenient tool. I use R for this.
Analyzing performance data
It is enough to have throughput (total, read or write), bandwidth (total, read or write), response time, read and write sizes and utilization for all SP, ports, LUNs and disks, to get a complex view of the performance.
We can see the total workload on the array, workload balance between SP and ports.
We can see the most busy LUNs and disks, and what pools and hosts they belongs to.
And we can classify components by performance metrics.
Now anybody can use it. All you need are CSV file with data and XML files with configuration and relations informations.
Let’s see the result – some screenshots.
On the next screen we can input form with instruction. After uploading all files we can see array name and S/N and the period data was taken on. We can select data for review.
A big problem for many production array is an understanding its configuration. What pools and RAID Groups we have, what LUNs are sits on its, what hosts uses whats LUNs, pools and RAID Groups, what disks used for its.
We can get configuration information in a visual view – as a graph:
And in the tables:
Some examples of performance metrics plots.
For classification we can select performance metrics and see the difference between the components.
And last but not least, the report. By pressing one button we can get a complete report in PDF format.
I hope this tool is usable and interesting for many EMC CLARiiON/VNX users.