The selection of a BI software package can be a difficult process to go through. There are alot of things to consider when making such a decision such as what your specific needs are as an organization, what skill sets your organization has available for support, hardware considerations, and cost. Here are a few ideas that might be useful when selecting your next Business Intelligence solution.
- Determine what hardware and operating systems the package is compatible with. Have the sales team document and lay this out up front, so you know if you have the skillset, software licenses, and hardware to use the package in the first place.
- Further layout what the additional costs will be for procuring software/hardware, purchasing support, and hiring or training a proper skill set. This will help you get to a total cost of ownership. Understand what the annual support fee structure is like.
- Make sure to walk through the development process. Often times, proof of concepts revolve around software capabilities, but it is just as important to look under the hood. A complicated development process can lead to many problems down the road for development and maintenance. Not all development processes are created equal.
- Every BI tool seems to come with a license for each function of the software. Licensing can get pretty complicated, so make sure you understand what each license covers. It really helps to lay it out in some form of spreadsheet, so you can total up how much licensing will cost based on what you select.
- Understand the capabilities of each package, simplify them into common terms across all packages, and attempt to compare these capabilities, apples to apples. Such capabilities might include OLAP, caching, scheduling, delivery, pixel perfect reporting, dashboarding capabilities, report development interface, user interface, architecting process, SQL generation, etc.
- Understand the technology behind the OLAP that is being delivered and determine if this will serve your organization adequately. Is the technology multi dimensional OLAP or relational OLAP? Multi dimensional OLAP performs faster, but also requires that data be stored physically on the platform. Relational OLAP may be slower, but sits directly on your EDW and usually is more flexible when expanding the scope of data that is to be reported. HOLAP which is a hybrid OLAP utilizes relational OLAP, but also allows for caching of data which can improve performance. Beware that caching may be sold as a multi-dimensional OLAP alternative, but there are limitations to caching’s effectiveness. In the past this was mainly limited by the RAM size of your BI server, but as technology advances, caching becomes more and more useful.
- Is it important to develop a proof of concept on your own data? It is definitely nice to see your data within a sales presentation, but building a POC directly on an EDW will most likely not be possible. Most vendors will ask for downloads of data that they will then store locally on their laptop for demo purposes. Because this is normally the case, the only real benefit of doing this is so that the business can see a presentation in terms they can understand. Don’t expect to gain any other type of insight about a product because they are using your data in the POC….unless they actually decide to develop on your EDW.
Well, that’s all I can think of for now. Cheers.
. I had to go to extent of allowing post to contain spam and I’ve found that to be the best ant-spam too for a forum based on 