Documenting the conflicting messages from Big Data London

My mission at the moment is to reinvent Kognitio’s documentation suite. Good documentation needs to be frictionless and inviting. It needs to inspire confidence.

Newcomers to Kognitio are at the start of their journey. At first, they might not even know what we do so an essential task is to present that message.

I thought it might be interesting to see how the exhibitors at Big Data London presented theirs.

The stands

I saw one or two familiar faces but many of the exhibitors were new to me, some only recently out of stealth mode. Looking at the wording on the stands, it wasn’t easy to guess what everyone had to offer.

Lots of them had ‘cloud’. Some had ‘NLP’ or ‘machine learning’. They had enough ‘analytics’ to fill your pockets. It goes without saying that the data was ‘big’ and the time was ‘real’. Of course, they had ‘at scale’. Some were ‘information driven’, some ‘compliant’. That sounded good. ‘Cognitive’ sounded even better.

But the big question was largely left unanswered:

“Yes, but what does it do?”

The fundamental communication of what a tech product offers is something most of us could pay more attention to:

  • You’ve worked hard to build your product.
  • You’re super-familiar with it.
  • Other people aren’t. They just aren’t.

It really is as simple as that.

One Data Scientist I spoke to summed it up perfectly:

“Putting ‘concurrent’ and ‘cloud based’ on your stand is like putting ‘hard working and punctual’ on your resume.”

The handouts

If the buzzwords on the stands were terse and slightly tired, maybe the message was in the handouts. The problem is no one wants to read 1,500 words to find out what a company builds.

Where’s the elevator pitch? Written down? In fewer than 150 words?

The real information started to reach me when I engaged with the staff on the stands. That’s when you start to find out about a product.

But this information really should be front and center in a form that’s almost effortless to digest.

And the talks

I attended four of the talks during my day at Olympia. My favorite moment came along by chance. No amount of planning could have arranged it.

I slipped quietly into the auditorium ahead of Kognitio’s presentation. The preceding session was drawing to a close. They were making a final point that Hadoop’s file system, HDFS, is for storage. The presenter emphasized that last word. The idea here is that HDFS is good at keeping your data safe but not great for heavy read/write activity. So there are some things you can’t expect to build on top of Hadoop with great success.

Some of the audience filed out but most remained to listen to Kognitio. I found an aisle seat next to a friendly, bearded giant. The Kognitio team took to the stage and began to mic up as the room filled to capacity again.

Sharon Kirkham and Chak Leung demonstrated a Kognitio on Hadoop system using data sampled from Transport for London. The dataset had all the bus routes in the capital. A hundred simulated clients fired off queries relentlessly, visualized as dots that flashed on a map so fast you couldn’t keep track. Real people don’t send back-to-back requests like that, so our 100 simulated clients probably represented 1,000++ real-world users.

The performance was blistering.

My neighbor glanced down at the Kognitio logo on my polo shirt, one eyebrow raised.

“Didn’t the last feller say you can’t do that?” he asked.

Leave a Reply

Your email address will not be published nor used for any other purpose. Required fields are marked *