For a long time, I was an Essbase guy. Like since before `my.yahoo.com`. I'm writing about how I came to understand the importants of DevOps as a kind of testimony about a culmination of the lessons of my prior positions. If you don't know, Essbase was voted one of the top 10 compute technologies of the 2000s by Information Age. And I quote:
The multi-dimensional database technology that put online analytical processing (OLAP) on the business intelligence map. Developed by Arbor Software (now part of Hyperion Solutions), it spurred the creation of scores of rival OLAP products – and billions of OLAP cubes.
There were two essential qualities to Essbase implementations that spoiled me and gave me a particular set of instincts that serve me now.
- Essbase was important, and it was disruptive.
- Essbase was always in production.
Let's talk about the second thing first. Essbase was built with the understanding that business rules were always going to be changing as business changed. So Essbase employed a semantic layer that was able to be edited in real time and then applied to its data models immediately. While it took some expertise to understand the batch time implications of any particular code, Essbase never supported a kind of migration framework to move its metadata from DEV to QA to PROD. This annoyed a lot of people, especially those fun folks who were the DBAs and ERP managers whose data was upstream from our Essbase cubes. Moreover, Essbase itself made the making of multidimensional models easier and cheaper than those built with traditional older tech. So the very idea of having multiple business models was part and parcel of this. So: multiple olap cubes, deployable in production with each of those cubes with multiple scenario dimensions and the facility supporting dynamic definitions of data-driven business rules. It was, quite frankly, more flexibility than most finance and planning organizations could wrap their heads around.
These days we call that kind of adaptability in production systems 'Continuous Deployment', which is a DevOps term. Now what Essbase did not have was Continuous Integration with systems outside of its aegis. But it did have tight semantic integration with its 'Essbase Ready' clients. Which meant that if master data changed in the database model, no additional changes had to be made to the reporting components. I am not aware of any next-gen databases that do that. That ability was part and parcel of the Essbase server engineers designing an API for their clients rather than just a pipe of text.
This combination of deployability and tight integration allowed us Essbase guys to grab data, get it into the model quickly and evolve the model right in front of our customers. That was why Essbase saved so many lost data warehouse projects. We surfaced changes quickly. We allowed our customers to explore data immediately and made multiple looks of flowing data happen on a daily basis, by design.
Essbase's importance and disruptive nature may sound subjective but it was not. I had the privilege of working for a hot Valley company that proved that it could IPO and make money. So when people decided to buy our database, it was a big deal and a high priority project. More often than not we were swapping out systems to the benefit of the financial guys and amidst the howls of IT. So when we in the field service group were assigned to get it up and running, the hot visibility was on us. Essbase lived up to the hype. In 3 years I never lost a trial.
The point is that when companies decided to buy Essbase, they went all in. It wasn't an experiment. It had to replace the company's financial reporting - all of it, or it was a failure. Essbase was originally not scalable to retail solutions, and that's one reason competing companies like Microstrategy were able to survive. But for the right fit, our customers swore by it. It changed the way people looked at IT because it just worked and it eliminated a large number of problems. These benefits were clearly presented up front before companies made their purchasing decisions. They bought the technology with the understanding that it would change their behavior, make business processes faster and improve transparency. It wasn't about having Essbase because all the cool kids had Essbase. It wasn't a market sweeping phenomenon. After all, we were competing in a crowded market. But for people who could see how this technology could improve their business, the way was clear.
I cannot say that what I learned technically using Essbase or any of the BI tools themselves helped me understand the broad variety of open source technologies that are associated with DevOps. Those were subjects that required their own mastery. But these two phenomenon having to do with the effect of the right technology on the business were instructive in my understanding of the conflicts and benefits that would arise upon implementation. I now know the effects of these changes in the organization and what kind of management approaches are necessary. In short, I have the experience of breaking ITIL standards and old habits of getting things done in medium and large organizations. Needless to say, you cannot simply drop technology into production and expect behaviors to change, but when you understand how functional behaviors need to change for the good of the business, it goes a long way in gaining acceptance to radical tech and process change.