Tag Archives: facebook

Google, Facebook Wouldn’t Have Been Possible Without Open Source

15 Feb

Red Hat is one of the leading companies operating on what is called the open source business model, where the source codes of software are open, enabling more programmers to make adaptations and improvements.

So, unlike proprietary software companies that do not open up the source codes and make money by selling expensive software licenses, Red Hat makes money by selling subscriptions for the support, training, and integration services that help customers in using open-source software.

Open source software has had significant success in many areas, and Red Hat itself has almost touched $1-billion in revenue. The company’s CEO Jim Whitehurst was recently in Bangalore and spoke to TOI.

What excites you right now in the open source world?
User-driven innovation. The Googles, the Yahoos, the Facebooks, who have significant IT challenges, are not looking to pay a vendor for intellectual property, they hire thousands of engineers to do it themselves, and they do it all on open source. They had ‘big data’ problems to solve, but did not have the dollars to solve them when they started off. That is why all the bigdata innovations, such as Hadoop and Cassandra (frameworks for running applications on large clusters built of commodity hardware), have happened on open source.

And now IBM, EMC, Oracle have all announced support around those open source projects. Why didn’t these things come out of the labs at IBM or Oracle? The simple reason is innovation is starting to move to user-driven innovation. It’s the explosion of Web 2.0 companies that’s driving open source.

Also, the pace of innovation at Web 2.0 companies is very rapid. The CEO of one of the big Web 2.0 companies I was talking to said he needed to launch a product in two weeks, so he had engineers working round the clock to ensure he met the deadline. Because it’s happening on open source, it opens up this huge, huge opportunity for Red Hat.

How cost effective is it?
One of the major banks in the US takes every single interaction they have with their customer and analyses it all in real time to take decisions. When they originally priced that system, it was going to be over a $100 million.

But eventually they built it with Hadoop, using Red Hat and commodity X86 servers for less than $2 million. So you are not talking about 20% savings, you are talking about radical savings.

The cost of building even the beginnings of Google, if you were doing that in the traditional IT stack way, would have been prohibitive. Our estimate is that if they had paid traditional license fees, even discounted, they would have paid $10 billion every year in such fees.

How is the cloud impacting you?
All major clouds are built on open source, other than (Microsoft) Azure. They either use ours (Red Hat) or free open source, and they use it right from the hypervisor to the operating system.

To build something on the scale of Amazon on a proprietary platform would be extremely expensive. So it does represent a huge opportunity. But the dollars are still in applications. Cloud as a multi-tenanted, off-premise set up is still a tiny, tiny part of IT infrastructure.

Software-as-a-service vendors are also building on our stack. Salesforce.com is built on Red Hat software. So we are in the cloud through that way too.

Do you believe everything will eventually move to open source?
This might be provocative for some people in open source, but no, not everything is moving to open source. Open source works well where there are broad communities of users, where there aren’t clear standards, or there is value in developing clear standards.

I’m willing to bet that IBM has profited a lot more from Linux than Red Hat has — by embracing it early on, offering it across their platforms. Sun Microsystems went the other way, and they almost died before they were acquired by Oracle. Open source is a very powerful developer model, very powerful economic model, but you have to figure out where to work with it.

Microsoft Windows still has the dominant share of server operating systems.

They are still about 60%. That’s because a lot of the applications still used are older ones. If you want to run Microsoft Exchange, you would use Windows. For customers to migrate applications is expensive.

But if it is a new application, we have a disproportionately high share. If you are doing only Web apps, or Java development or any new development, that is disproportionately Linux. In countries like India, where it’s significant new IT infrastructure, we have great opportunity.

You recently acquired Gluster (it has almost its entire engineering team in Bangalore). Where does it fit in?
There’s today an explosion of data, especially of unstructured data. How do you store and manage it? The people working on this problem are primarily the storage companies like EMC,NetApp. These are hardware companies, so for them the solution is a box that sits on the floor in a data centre. The thing that’s great about Gluster is that it’s a software solution to the problem.

You can plug up commodity disk drives to this thing, and you are ready for the application. It looks like one big disk drive. More importantly, because it’s software, you can spool it right up into the cloud. Pandora, the music streaming service, is a customer.

If a million kids show up to listen to a song, they can’t all hit one disk drive, so what this file system does is auto-replicate that song may be 50,000 times so that each one is hit by only 20 people. It can also spool right up on Amazon. When a song goes viral, it can spool up, and when the demand goes down, it can spool down. You can’t do that with a hardware solution.

It’s an open source solution, it’s low cost and modular. Red Hat started with the operating system, then middleware, virtualisation, and now we also have storage. Gluster has 50-60 people, we’ll likely double that this year.

Source: THE TIMES OF INDIA

Advertisements