An Information Management interview with Steve Miller, co-founder of services/outsourcing firm OpenBI
How has open source worked out for your consultancy since you launched OpenBI four years ago?
If anything, we might have gotten in a little early, though we've always been profitable. The industry might not have done as well as vendors want you to believe. Some of the growth numbers they provide have come over pretty poor performance, but I think they're heading toward profitability now. There was some naïveté that we'd come in and start taking business from Business Objects and Cognos and so forth, but that's only happened a little bit. After 2007, we started to see a nicely sloped increase in the use of open source business intelligence. I don't know if there's going to be a “crossing the chasm” moment, but I do see steady growth of this technology.
What has changed in the last couple of years?
In the beginning, we were consulting almost exclusively with OEMs and ISVs. Their needs are relatively small, and we were hoping for more enterprise customers. Now we are seeing more enterprise customers, no question, and we take that as a positive. Those customers tend to do a lot of projects. We get on board for two or three months, and then there's another project and another after that. You know, it's those aha! moments in BI that lead to more questions. Plus, there's more acceptance now. We don't have the big open source religious arguments that we did four years ago.
What's the attraction of open source for your enterprise customers? Flexibility? Cost?
Where it's technology people driving the decision, it's more about flexibility and the open source technology mantra. But often as not, we're brought in now by a line of business that needs to put together an application inexpensively.
What do the projects and their scope look like?
It really depends on how much help they need. Our typical projects are scoped at six to 10 weeks. It's often to build a data store, conceptualize and design a database, do the ETL [extract, transform and load] and reports, dashboards and OLAP [online analytical processing] cubes. There are caveats in the number of transformations and reports and other things we provide. We often get into horse-trading, depending on the client's needs and budget. We tell them what we think we can do in a time frame and let them interpret that from their point of view. Some will say, "I don't want dashboards or OLAP, I only want reports." So we trade two reports for half a dashboard or something like that. But as we come through with those deliverables, our hope is that customers are excited and want additional projects so we end up with a series of those kinds of projects.
Are clients more aware of what they want and need when they start working with you?
Some clients are quite naïve as to what they're getting into. We run into customers that don't understand how intensive a data exercise business intelligence is, open source or otherwise. They're thinking, “I want a few reports for this, and a few dashboards for that.” But they're oftentimes not thinking about the foundation and infrastructure you need for a BI environment - that requires an education. I would say that in excess of 50 percent of our work is data movement, ETL, data cleansing kinds of activities, If you ask most customers how much work is data and how much is the app, they would probably say it's 15 percent data and 85 percent app. It usually turns out to be quite the opposite.
How do you see the future of commercial open source vendors?
I would say that commercial open source is already something of a misnomer. You're paying for a subscription to the software. There is a "freemium" or open core model where there is a free community product. But it's not terribly functional, and more and more, the vendors are putting the secret sauce into proprietary components. It's not surprising, vendors initially thought that providing support would be the way to go, but that hasn't delivered what they thought. Commercial open source vendors are starting to look more and more like proprietary vendors, the difference being that you don't license the software, you subscribe to it. You pay your annual fee, which is kind of in line with what a maintenance fee from a proprietary vendor would look like, but you're not paying the huge initial licensing cost.
That still sounds like a service model. So does open source meet up with software as a service in that regard?
We think that open source is a great foundation and platform for SaaS. We have already done a few cloud deployments, and we're in discussions with several ISVs [independent software vendors] that are software as a service, multi-tenancy kinds of products looking to use open source for that.
Have scoped projects and the service model changed consulting generally?
Absolutely, there's no question. Ten years ago, we wouldn't have looked at a 10-week project - that would have been on the very low end. It's a different world today, but there are still great opportunities. You just need to keep a lineup of those, and we've had success with that.