Q&A: Weighing Pros, Cons of Open Source BI
Open source BI solutions have matured; now they offer sound and compelling software for business intelligence, but they may be overlooked in the search for a solution.
- By Linda L. Briggs
- March 11, 2014
Companies considering open source solutions for business intelligence need to weigh the same considerations as with any software solution, but there are additional considerations as well. In her book Using Open Source Platforms for Business Intelligence: Avoid Pitfalls and Maximize ROI, analyst and author Lyndsay Wise explores the benefits and challenges of selecting an open source BI solution. In this interview, she touches on some of the key points covered in the book.
Wise has over 10 years of IT experience in business systems analysis, software selection, and implementation of enterprise applications, She is president and founder of WiseAnalytics, a company that focuses on small and midsize companies and conducts research into leading technologies, market trends, BI products and vendors, mid-market needs, and data visualization.
BI This Week: Why a book on open source and BI?
Lyndsay Wise: I focus my business mostly on small and midsize businesses, so I know that lots of organizations don't have the time or resources to weigh and implement a large-scale business intelligence offering. The book grew out of a desire to explore alternative solutions.
Also, many companies may have a few developers on board who are already open source enthusiasts because they like to develop things on their own, then customize them to the organization's needs. However, those developers may not know BI at all. Open source is an area lots of organizations don't even know is an option in BI, or don't know whether it's right for them. The book is an exploration of alternatives available to companies that may not know what's out there.
So open source is often overlooked as a potential BI solution?
Yes. I find that organizations often have open-source enthusiasts, but they don't necessarily identify the benefits of selecting open source for business intelligence specifically. On the other hand, they may be such open source enthusiasts that they think it's great for everything -- but it might not be the best choice for BI.
What has changed in the past five or more years to make open source more viable as a BI solution?
Before I thought of writing a book on open source, I was speaking with someone with an open source vendor who said that when they originally attended open source conferences, they had to educate the market about what BI was. When they started moving into the BI realm and marketing themselves alongside other BI vendors, [the conversation became], what do open source solutions offer? In either case, it was all about trying to educate the market -- that business intelligence is broader than you think and that you can apply open source to BI.
Does open source tend to be a departmental choice or a small or midsize company solution? Why is that -- or is that changing?
I think that is changing. That was my viewpoint when I started writing the book, but as I started speaking to vendors and studying case studies, I saw that open source tended to be a choice in larger companies and for broader deployments simply because those companies have the resources on hand to develop solutions without taking away from other development activities. There are lots of large organizations choosing open source.
At the same time, open source BI is starting to shift as vendors offer different tiers of solutions, both commercial and community. The commercialization of these solutions, which offers better support and better user interfaces, lets many organizations implement at the departmental level or on a smaller scale.
Speaking of that -- and you cover this in an entire chapter in your book -- can you describe the levels of open source BI? Which option is best for what kind of company?
I like to look at it as having three levels. The first is traditional community open source, where someone uses the source code to develop their own solution. In organizations making that choice, it's really important to have someone who is familiar with open source. To be effective, it's best if you already have some expertise in the area that you're developing in. As I said, that approach usually works for either large organizations, smaller organizations, or departmental solutions that can support developers who are already in-house.
The second type of open source -- and this is where a lot of traditional open source solutions started -- is one in which the customer base not only offers their own solutions but they offer support. I'd call that community with support.
The third type of open source is for organizations that don't have developers in-house, don't have an internal IT department, or have an IT department that is largely focused on support. If those organizations want to select open source, it's probably best to go with a commercial version. That means it can be supported, just like any other product, whether using professional services internally or consultants external to the company. In either case, you have support over time until your staff is well-versed in the product.
What's confusing the market now is that many open source vendors aren't marketing themselves as open source any more. They really want to be competitive within the data warehousing and business intelligence and data integration markets, so they often won't talk about their open source differentiator. That becomes an issue when a company is trying to learn about different solutions and maybe doesn't realize that the vendor is describing their commercial version, not their community version.
It sounds like vendors may shy away from the open source designation for their products.
Yes, it's interesting. One of the reasons open source was regarded in such a positive light initially was because of a constant focus on continuing to make tools better and keeping them open rather than making proprietary software that you can't really turn into what you want. However, because the BI market is a little different,many of vendors try to downplay that aspect. In doing that, however, there ends up being a bit of a disconnect between the benefits that an organization can get by using the community version versus the commercial version.
Let's talk about some of those benefits. Open source is often cited for low cost considerations, but what are some other advantages?
Cost is definitely a factor. Another advantage is this: When your organization wants to get something up and running quickly, to immediately show the value, and to be able to customize BI software quickly and get a working prototype -- that's a great time to use open source. ...
Is it a common scenario to start with one level of open source and move to another?
It definitely is. Lots of companies want something that is low risk, and either they've tried a BI solution in the past that hasn't worked out or they don't necessarily want to deploy a large solution until they know that something has worked, whether it's for one department or for one function.
In those cases, they may use the free version first, then move up not only to get support but to get some of the functions available only with the commercial version. That's a great approach, by the way -- it's tried and true, the integration is there, and companies can expand their solution as they go.
Licensing seems to come up as a concern in OS discussions. I know you cover that extensively in your book -- can you recap some of the issues and considerations here?
Licensing extends beyond open source. I find that vendors are becoming much more creative, and better, at their licensing options. One thing to consider is that lots of open source vendors use more of a subscription-based licensing plan, which is similar to software-as-a-service vendors.
At the same time, there are all different kinds of licensing, depending on the specific vendor. One common choice is developer licensing versus user licensing. It depends on the vendor. Some offer named licenses, where only certain people can use them. Other plans depend on how many people are using the software at the same time. In any case, you'll want to carefully weigh your current use versus your growth plans.
In general, vendors are becoming much broader about the types of licensing agreements they offer. That's because the traditional models often don't work if you want to scale up and out. That's true of all vendors, by the way, not just open source. ...
It's really important to do your due diligence. That's the base -- really understanding the licensing models that are offered. Look hard at the implications if you expand your adoption of the product -- whether that means the number of developers using it, the number of users, how it's being used, or how it's being deployed. Consider what any changes will mean for your actual licensing costs. You don't want something that works now, but once you need to expand, it doesn't necessarily fit your needs anymore because the scalability just isn't there.
What things should a business manager consider before going with an open source BI solution? What about a VP of Tech or CTO? Are the considerations different?
When it comes to the business side of things, it's important to do due diligence, as I've said, whether it's an open source solution or not. What I've found when I've spoken with different organizations is that there are business managers who are interested in open source because they have either developers or IT staff internally who are very enthusiastic about open source simply because they've used it in the past. Instead of doing their own due diligence, business managers can rely too much on other people's historical knowledge -- and that might not necessarily be specific to BI. Because an IT person might not know how open source is going to be used in BI, it's important to do your own research and homework and not just rely on IT.
Conversely, on the IT side, if an organization has been using open source and has the internal resources, and the solution can benefit business, then open source is a good choice. However, as with any technology solution, it definitely depends on the business requirements. You want to look closely at what's currently being used internally and who exactly can develop the open source project. There are developers who will say, "Sure, we can develop that," but they might already be working on three or four other projects. It's important to be realistic.
Where's the ROI in open source BI, and how much of a role do cost considerations play in open source BI?
At the beginning, when open source was starting to become more popular, cost was a huge consideration. As time has gone on, and you have more vendors offering a diverse range of free products, free versions, free trials, and so forth -- such as Tableau or MicroStrategy -- I don't think the "free" aspect offers as much pull as it used to. However, using open source can certainly be quicker.
Where open source might become more popular in certain circles is with big data -- with Hadoop and MapReduce and all of those open source platforms. ... Those products may bring more attention to open source again. I haven't done any studies on open source adoption and big data, but I think it definitely has opened up awareness of open source in general and its benefits.
Your book is 200-plus pages, so I know there's plenty we haven't covered here. Is there anything in particular you wanted to touch on?
I started this book thinking very positively about open source, and I still do see it in a positive light. However, the more research I did, and the more writing I did, the more I realized that open source is not for everybody. Any organization can choose the wrong solution -- even if something costs nothing, as open source can, time and resources are money.
It's a book about open source, but it's important to realize that it's also a book about choosing the right BI solution in general. Open source is just one type of business intelligence solution, so if an organization is looking for a solution and wants to try to figure out what is best, I don't necessarily look at it as open-source versus non-open-source choice. Instead, look at what's best for the organization now, and the potential for scalability. In some cases, the answer will be open source, and in some cases it won't.