Today I start building my massive server in earnest. As we speak I'm dowloading a big hunk of software from Oracle. They were very smart to do some of what Hyperion did which was to put all of their supported versions in one place so the customer base could get at it. But they went one further which was to rely on their legal teams and not their DRM to control distributions. So I've got access to the entire stack of software which I can use as a developer and not worry about calling 15 different people to negotiate a developer's license. I'm glad that era is over.
That doesn't change the fact that it takes time to assemble all of the right pieces and installation and certification of the installation is a bear. So that's where I am now. I'm getting OBIEE, Essbase, Informatica, Analyzer and Warehouse Builder. That should allow me to recreate, as is my intent, a big library of my greatest hits. Which is to say, I'm going to, as time permits, put together the biggest collection of demo apps the space has ever seen (if I say so myself). See I've got over 100 different customer histories in Vault 107, and I intend to strip them down, very carefully so as not to violate that sacred trust between developer and customer, and remake the salient pieces of their applications into a thing of complex beauty. Half of my task is to pick 20 from the 150 odd and prioritize. But that's just half the fun.
I remember doing this the first time when I moved from Pilot Software over to Arbor back in 1996. I was amazed at how easy it had become to build what had taken months in just a matter of days. So if these products are all that, I should be able to build in a matter of weeks, the best of what I've been doing the past 11 years. What's going to help is that I'm going to use my new Java skills (fingers crossed) to flesh out a multidimensional data generation tool whose core code I've already written in Perl. With any luck, I'll be able to figure out the API pieces and get it to read an Essbase outline. So this is my private project for 2009, which is to build a rather sophisticated data generator that will read your outline, and generate fake data so you can do performance testing. So far as I know, no such tool exists.
What I really want is a database scale visualizer - something that parses the Essbase index and then gives a navigable kind of heat map or something which allows developers to quickly asses what data is doing in cubes. For example, wouldn't it be cool to have a visualizer that allows you to see a calc as it's processing? Something like that. I know that it's stupid ambitious, but hey. I'm getting bored with politics and freeing up my time to do technical things.
In the meantime, the installation continues.
Scapa does something similar, it generalizes a query based on the outline based on an initial query, but doesnt generate data. Then you can used the generalizations to hammer the cube with queries.
Posted by: Rohit Amarnath | January 14, 2009 at 05:40 PM