Big Data Makes Organizations Smarter, but Open Data Makes Them Richer
Big Data, meet Big Hardware.
Enterprises looking at a tidal wave of massive data volumes with the advent of digital video, social networking and massive database files are seeing new means to keep their heads above water. But questions are starting to surface as to whether large, expensive hardware platforms are really the best way to approach the problem.
IBM made significant strides against Big Data this week with a revamped PureSystems platform kitted out with an advanced analytics engine geared to process petabytes of data within minutes. The company says this will enable enterprises to break down key business processes, such as marketing, sales and business operations and determine where and how they can be improved. The PureData System comes in three flavors: a transactional model designed to handle high volumes of extremely small data sets; an analytics model aimed at data warehousing and predictive analysis; and an operational analytics model that hones in on tasks like fraud detection and real-time change management. IBM also says the system is built on a streamlined hardware platform that can be deployed and operational within 24 hours.
As well, Cisco Systems has teamed up with a company called Pentaho to develop an integrated, open-source platform designed to support Big Data and business analytics. The package combines Pentaho’s analysis and reporting software with the Cisco Unified Computing System (UCS), leveraging such hardware systems as the M3 server and various networking fabric technologies. The system is designed to run analytics tasks on raw iron or in virtual/cloud environments. The tie-up suggests that the vendor community, at least, sees integrated hardware/software platforms as the best way to handle the unique challenges Big Data presents.
But is that necessarily so? According to GigaOm’s Stacey Higginbotham, x86 commodity hardware would probably be too expensive to provision and operate to handle Big Data in parallel, but low-power cores from SeaMicro, Caldexa and others could probably get the job done. The question enterprises have to ask themselves is whether such a system is worth the architectural challenge, or will the plug-and-play aspects of a specialized platform win out on simplicity grounds? And in the end, do these platforms represent hardware vendors’ admissions that services really are the profit centers of the future?
In fact, if you really want to gain true insight from all that Big Data, you don’t need expensive hardware or software, according to McKinsey and Company’s Matt Ariker — all you need is a pencil. So few organizations even know what they are looking to glean from their institutional knowledge that many end up spending oodles on systems that they ultimately have little or no use for. To help guide the process, he recommends CIOs write down answers to a few simple questions, like “What business impact am I looking to make with Big Data analytics?” and “How will I know what success looks like?” By giving it just a little thought up front, enterprise executives will have a better handle on provisioning for what they need, rather than what they think they want.
No doubt, Big Data is a big challenge, but big challenges are often more adequately addressed by intelligent thinking than by brute force. The new integrated platforms profess to employ both at a time when enterprises are under the gun to produce streamlined, yet highly effective, solutions to the problem of increasing data loads.
In that regard, perhaps it’s better to view Big Data not as a problem of size, but of complexity.
