Next Generation Outline Extractor: if you don't have it, get it

For what ever the reason is, I often find myself in need of some sort of outline extract, be it a flattened generational view, level view, a load file, or whatever the case is, you just can’t easily pull it from Essbase all the time. It’s not that Essbase does not allow you to pull an extract of dimensions; it’s just not always the format you are looking for. This is where the Outline Extractor comes into play, this little tool is fantastic and if you are like me, it’s going to make your life much easier.  

I’ve been using the Outline Extractor for a couple of years now, and at KSCOPE16 I had a chance to attend a session by Tim Tow, presenting the latest and greatest iteration of the tool, Next Generation Outline Extractor, and one feature in particular caught my attention.  The Outline Extractor now has an option to ‘Extract and process MaxL Outline XML’ instead of using the Essbase API.  What this means is in some environments, particularly high latency ones, there are ASTRONOMICAL gains in speed.  To quantify this I did a case study.

Using one of my production applications, running Essbase 11.1.2.3.500 on an Exalytics box I decided to extract a dimension using both methods to compare. I'm running in an environment where my latency to the server is between 230 and 260ms, and bandwidth is not 'great', so using the Essbase API was a bit time consuming.  The dimensions I am extracting has 4829 members, and I'm using all default options in the tool, and outputting in Generation, level and Load File Formats.  So with that let’s take a look at some quick numbers.

Essbase API – extraction 242m 26.29s

Extract and process MaxL Outline XML - extraction in 0m 13.76s

Yes you read that right, there was no typo there.. I did the exact same output for both, and those really are the result, that’s an improvement of over 100,000%. I don’t know how to better quantify this, all I can do is say use this option, use it often, and don’t look back. Just to make sure it’s clear, here is a screenshot of the option you want to select

Now that using outline extractor has become a quick job requiring only a few seconds, let’s take a look at why I love this tool so much.  I'm not going to put together a usage manual for the tool, Applied OLAP does a great job of that in their documentation, and they have great support if you get stuck, what I want to show is the outputs you can get out of this tool.

The tool can be run from a command line, or a GUI utility, the GUI just creates the .properties file that you could create manually and run with command line, but I really like the GUI, so I have spent most of my time in that mode. It will take you through the options step by step, including a very nice ‘cube browser’ to select exactly what cube you are going to be extracting from and a very easy to use Dimension selector, where you can even apply include/exclude filters

The next screen is where you can choose one of the many available formats for the output, I personally have not worked with the Relational Cache Writer, but I’ve heard good things. The 3 you can see selected here are the ones I use the most.

There are lots of options, so you are sure to be able to format the file exactly how you need:

Select your performance options next, I normally leave this as default, but if you want information on these, I would recommend checking out the products included documentation. After that it’s time to run, save your new properties file and execute! All of the log information will be show on screen so you can track progress and you are all set.  Take your extracted ‘Load File’ and load it into another cube if that’s what you need, if business users are looking for a generational view, you can pass that along all with almost no real time investment from you.

One last point, did I mention this is all free? Yes, that’s right, it's a free tool, and a good one at that, thank you Applied OLAP, it’s more than a little appreciated here.

I really hope that this will help you as much as it helped me, if you have any questions please leave a comment.