C-level briefing: CBR spoke to Arun Murthy, founder of Hortonworks about Hadoop trends, Spark and not pushing new technology.
The speed at which the Big Data market has grown will have left many heads spinning as they seek to catch up with product releases and trends.
Hortonworks and the Hadoop community have recognised this:"It’s something we recognise both as a community and a vendor. We are probably now far less aggressive than we were a couple of years ago.
"We also understand that we could put another 25 technologies into our platform at this point, but I think we and the customers would be better served by taking the core of our technologies we have and making them easier to consume, easier to use, monitor, deploy and debug," said Murthy.
This is something that he says the Hadoop vendors are reacting to, resulting in them not pushing new technology as much as they had before.
"I think you are getting a lot more attention to fit and finish rather than to just getting the new technology in," he said.
The company is making sure it ships stable versions of the platform, as the reaction from a customer to an unstable technology could be catastrophic.
"If people have a bad experience with it then they will never touch that piece of technology again, so we are trying to be really careful about that."
Identifying trends in the Hadoop market, Murthy sees newer use cases on the platform being vectored towards Spark. One particular area is predictive analytics – which he says Hadoop is well suited to.
"Predictive analytics is an over-used term, but I think the difference with something like the Hadoop platform, is that you can actually do it at a very fine grain.
"You can optimise and model every single user rather than trying to collapse them into a segment or a cluster.
"Doing that really helps because now you can actually whether it’s targeting products or healthcare or any of these things.
"That I think is the difference between predictive analytics you can do on Hadoop compared to a legacy platform."
A growing trend with Hadoop is the growing interaction between businesses and the open source community. More frequently Murthy is seeing employees become contributors to the open source project.
This is something he wants to see more of as it has the potential to broaden the appeal of the software.
"Now when the software comes out all these guys know it’s ready for them because they actually invested their time and effort into it."
This means that businesses are shaping the technology to their needs rather than getting a big vendor to do it for them, which has its downsides.
Murthy said: "In the past they would go to a big vendor give them some requirements, two years later you’d see they asked for a horse and they’d get a unicorn."
Looking forward Murthy hopes to see more out of the box products, something that will simplify and grow deployments.
"Right now everyone has to implement/write their own Spark logic, at some point that’s not a sustainable thing. There are only so many people who understand Hadoop then Scala then Spark; it gets tougher and tougher as you go down the chain.
"I’m hoping to see more built out products, ISV’s picking up Spark and other tech, to build products."
As data becomes possibly the most important asset that a business has, the simplification of tools and services becomes more important.
Allowing businesses to catch up by not constantly throwing new technology at them is one way that Hortonworks and Hadoop is helping. It helps the business to get to grips with this important technology.