Here's a real-world example:
A major Internet banking platform was set up so that most alarms would get triggered when systems reached static thresholds like 80% utilization. Advanced analytics were checking on the system in this case too. The analytics software noticed that a CPU was showing 15% utilization when it should be at 5%. Something was amiss. But it wasn't a crisis yet, so the admins said, "Well, there's still 85% to go."
At about 4 p.m., a whole bunch of users started getting off work and trying to deposit their checks and withdraw money at the end of the day. At that point, a request that had been stuck in a loop finally kicked loose. And a whole day's worth of transactions that had been bottled up behind that clog in the system all came flooding in at once.
Think of it like a garden hose that had been clogged, gradually building up pressure all day, bulging and swelling like in a cartoon and then just blowing through like an uncontrollable fire hose. The system was down for critical hours, right at the end of the day, when the users were counting on it the most.
Real-time analytics can give you an early warning that you have a kink in the hose and help you locate it and straighten it out, before it washes up everything downstream.
Does ITOA sound like it's too good to be true? Maybe. But Gartner predicts the sector will grow from essentially zero a couple of years ago to over $2 billion in 2018 global sales. We are also starting to see and hear about more and more real-world use cases. A number of us in the industry have been working on the problem and we are finally seeing the kind of compelling results we've all been looking for.
The three things that you should do now
* Research. There are a lot of analysts and companies parroting a lot of the right phrases, but who don't really understand this market yet. They are learning too. Be careful! Take your time. Learn the space. Research the researchers and the software or service providers. Learn what can really be done with Big Data analytics today, and start talking with your own internal teams about how it can apply to your business processes.
* Test drive a real use case. After you've done some research, figure out a few real technical or business problems to address. You'll want multiple use cases because different vendors may be able to help with different problems. The technology may be ready to handle some problems and not others. Just like different screw drivers in the toolbox can't all be used for all jobs, the same applies here with analytics tools.
Sign up for MIS Asia eNewsletters.