Dr. Dobb's is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them. Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.


Channels ▼
RSS

Extracting Symmetrix Data with Orca


Extracting Symmetrix Data with Orca

Tom Kranz

The Symmetrix is EMC's flagship storage device. It's big. It's bad. And it can sometimes be a bit opaque, meaning that performance data can be difficult to extract. EMC produces a range of tools -- some graphical, and some that can be called from the command line -- but using these to generate long-term performance statistics, in a management-friendly format, can be a bit of a struggle.

Orca is an open source package by Blair Zajac that uses RRDTool to plot arbitrary data from text files. Traditionally, Orca (http://www.orcaware.com/orca) has been used in conjunction with the SymbEL Toolkit to plot Solaris performance data. In this article, I will show how to use EMC's command-line tools, in conjunction with Orca, to plot performance statistics for your Symmetrix.

For purposes of this article, I'm assuming that you have rsync, ssh, and some sort of Web server installed -- these should form part of every sys admin's toolbox. You'll also need SymmCLI installed on the host connected to the Symmetrix (I have version 4.1). You'll also need Orca already installed and configured; I highly recommend it merely for the out-of-the-box Solaris performance monitoring it gives.

About Orca

A process (called orcallator) runs on a host and logs data into a file. Using rsync, this data file can then be copied to a central server, where Orca is run to process the data. Orca places that data into RRDtool databases, and then generates graphs from those databases. Orca works out averages and percentage increases by using the date stamps, which are written to each line of data in the orcallator output file. The graphs and output files are then placed in a directory where your Web server can see them, and voilà! Your performance data is available via a Web browser. See Listing 1.

Collecting Data from Symmetrix

The trick here is to pull performance data from the Symmetrix and then log that in a file that Orca can understand. symmstat is the SymmCLI tool that queries the Symmetrix and returns performance statistics. Listing 2 (orca-symmetrix_probe.pl) is a Perl script that calls symmstat, parses the output, and then generates a data file that Orca can understand.

We'll collect all the performance output from symmstat -- it's all useful, and is best looked at all together rather than focusing on one statistic (e.g., writes per second).

symmstat will provide the following data:

  • I/O reads per second
  • I/O writes per second (Figure 1)
  • kb read per second (Figure 2)
  • kb written per second
  • cache reads per second
  • cache writes per second (Figure 3)
  • sequential reads per second

A normal I/O read is when one block of data is requested and received. A sequential read is when we request several blocks of data, but they are all next to each other. The disk head can position itself, then just read several blocks, one after the other, which is an efficient way to get data off the disks. Write pending (wp) tracks (Figure 4) the number of tracks that the Symmetrix is sitting on; it's waiting for a drop in I/O before it stages them out of cache and down to disk.

One of the big advantages of getting Orca to graph these figures is that we can easily look at them together, over time. For instance, a high number of read I/Os, combined with a large amount of kilobits being read, with high numbers of sequential reads, could be the normal result of large table scans in your Oracle database (which might not be a problem).

A large number of small writes that aren't hitting the cache (large number of I/O writes, small value for kb writes, and small number of cache writes) will mean that your overall I/O performance will be suffering, with the Symmetrix constantly having to write to disk. You'd need to look at what sort of I/O is taking place, or maybe invest in some more cache.

Note that you can use this information to get an I/O "footprint" of your applications and then, using the historic graphing of Orca, you get an idea of when things are ok, and when they are not. You can also use the Orca graphs for capacity planning or gauging what impact a new software release will have.

It's worthwhile keeping an eye on the amount of write pending tracks. Compare it to the amount and size of the write I/Os occurring. Ideally, you want the Symmetrix to be busy, but not so busy that it gets overwhelmed.

When pulling the data from the Symmetrix, there are two main things to remember:

1. We need the epoch date (Orca uses this when working out averages over time).

2. If symmstat returns nothing for a device (which it can do regularly), we need to trap this and put a 0 for that value in the output file.

orca-symmetrix_probe.pl (Listing 2) takes symmstat data and creates Orca-friendly output files. It should be called from root's crontab every 5 minutes. Once it's running, you should notice the output file being generated and updated in /var/log/probe.

I have been using this script against a Symmetrix 3630. If you have a different breed of Symmetrix, you might notice that symmstat reports slightly different information. In that case, you'll need to edit the script to suit.

Once we're gathering data, we need a mechanism to copy those output files back to the central Orca machine for processing. Listing 3 (orca-emc_getlogs.ksh) should be run from the Orca user's crontab every 30 minutes. Be sure its execution doesn't coincide with the orca-symmetrix_probe.pl script (Listing 2), otherwise you'll be copying an incomplete data file.

Essentially, this script calls rsync to copy across the data file, and then calls orca to process that data. I'm using rsync over ssh because of the secure password-less connection that ssh offers, and because it ensures we only copy updated data, vastly reducing network traffic.

When we've called Orca from the script, we're passing it the enclosed config file, which tells Orca what values to take from the data files, and how to plot them. With this setup, you should now be able to launch a Web browser and view the performance graphs for your Symmetrix.

Whether the information you gather is sufficient to help you secure an extra gigabyte of cache is another matter, but at least you know how the Symmetrix is performing.

My thanks to Mick Sheppard for his invaluable help with the intricacies of Perl.

Tom Kranz is a contract sys admin with more than 7 years of experience. His main skills lie with Solaris, IRIX, and SANs. When he's not stuck in front of his Octane, he can be found herding his children round in a classic Rover.


Related Reading


More Insights






Currently we allow the following HTML tags in comments:

Single tags

These tags can be used alone and don't need an ending tag.

<br> Defines a single line break

<hr> Defines a horizontal line

Matching tags

These require an ending tag - e.g. <i>italic text</i>

<a> Defines an anchor

<b> Defines bold text

<big> Defines big text

<blockquote> Defines a long quotation

<caption> Defines a table caption

<cite> Defines a citation

<code> Defines computer code text

<em> Defines emphasized text

<fieldset> Defines a border around elements in a form

<h1> This is heading 1

<h2> This is heading 2

<h3> This is heading 3

<h4> This is heading 4

<h5> This is heading 5

<h6> This is heading 6

<i> Defines italic text

<p> Defines a paragraph

<pre> Defines preformatted text

<q> Defines a short quotation

<samp> Defines sample computer code text

<small> Defines small text

<span> Defines a section in a document

<s> Defines strikethrough text

<strike> Defines strikethrough text

<strong> Defines strong text

<sub> Defines subscripted text

<sup> Defines superscripted text

<u> Defines underlined text

Dr. Dobb's encourages readers to engage in spirited, healthy debate, including taking us to task. However, Dr. Dobb's moderates all comments posted to our site, and reserves the right to modify or remove any content that it determines to be derogatory, offensive, inflammatory, vulgar, irrelevant/off-topic, racist or obvious marketing or spam. Dr. Dobb's further reserves the right to disable the profile of any commenter participating in said activities.

 
Disqus Tips To upload an avatar photo, first complete your Disqus profile. | View the list of supported HTML tags you can use to style comments. | Please read our commenting policy.