How To Run An Effective Big Data POC

Posted by: Chris Neumann Posted date: May 16, 2011 In: Aster Data, Big Data | comment : 0

Ahh, the POC. It’s like test-driving a car, except the salesperson is doing everything in their power to keep you from touching the wheel.

In my previous post, I discussed the fact that in today’s market, there isn’t a single Big Data vendor whose product does everything. Each vendor has their strengths, and each vendor has their weaknesses. With sales reps doing everything in their power to keep you from discovering their dirty little secrets, how can you determine which solution is truly the best for your needs?

First off, before any of the vendors come in the door, it’s essential that you consider the factors that will influence your decision. Obviously, everyone wants the fastest/cheapest/best solution, but what else is important?

Here are a few things to consider internally before you get started:

1. What are the mission critical use cases?

Before you evaluate potential solutions, you need to decide what use cases you’re going to use to compare them. The obvious starting point is the set of use cases which are driving the purchase (the performance and/or functionality challenges which originally triggered the POC). However, a common mistake is to limit the list of POC use cases to these “pain points.” The problem with this approach is that you may end up choosing a solution that alleviates your current challenges, but introduces bottlenecks in other areas. When preparing the use cases for vendors, be sure to include both top-of-mind challenges and mission critical use cases (particularly those that run fine today, but which you can ill afford to negatively impact).

You should also spend time to rank these internally in order of importance, both in case you don’t have sufficient time to run them all, and so that you know how to rank the vendors when the results come in (though definitely don’t share that list with the vendors!).