When to Build, When to Buy, When to Outsource - Data Just Right: Introduction to Large-Scale Data and Analytics - page 179

13
When to Build, When to Buy,
When to Outsource
T hroughout this topic, we've explored the best choices of technologies for a variety
of use cases. We've taken a look at systems that collect data at a rapid pace and scale
up accordingly. We have covered strategies for collecting large amounts of data in real
time and for asking questions about this data quickly. We've even looked at ways to
store and share huge amounts of data.
As with any highly innovative and cutting-edge technology, the range of software
for dealing with data challenges exists in a variety of states for both development and
adoption. This places people who are trying to solve data problems in a bind. Does the
solution require investment in hardware administration and software development, or
does the answer mean purchasing services from a commercial data-solution vendor?
There is no universally correct answer to this question, but we can look at com-
mon patterns to help inform us of the right choices. In this chapter, we will take our
earlier principles of choosing just the right data technology—understanding the use
case, planning for scale, and avoiding managing infrastructure when possible—to help
understand when to buy versus when to build.
Overlapping Solutions
Here we go again! With each new technology revolution, an age-old IT problem rears
its head: Should we build our own solution, or should we buy an existing product?
The build versus buy conundrum appears so often that a whole range of consultants
and methodologies exist to inform decision makers about the best course of action.
For those working with new technologies that fall under the Big Data umbrella, this
question is coming up more and more often. Although open-source software has
often been the driver of new innovations in data technology, more and more com-
mercial products are appearing. Simultaneously, the growth of cloud computing has
also provided the ability to use hosted virtual servers instead of traditional server- or
appliance-based solutions. Unlike other technology cycles, in the data world there's an