It's easier than ever to scale like Google without being Google

May 07, 2015



3 min read

"Google is an infrastructure company with a great UI," joked Urs Hölzle, the company's senior vice president of technical infrastructure, during a talk at the Collision conference this week in Las Vegas.
It's not just a great soundbite, it's also partially true. In the past decade or so, Google has created dozens of infrastructure technologies in order to ensure its applications continue to improve even as they grow in number and continue to deal with more data and more users.
However, Google wasn't always in the business of creating its own technology. In part, this was because the company wasn't yet the behemoth it is today. But it was also in part because Google didn't have the resources do all the things it needed to do.
Until about 2003, Hölzle noted during the talk, "we were pretty much in survival mode." If a technology investment didn't result in immediate payback, it probably wasn't going to happen.
While Hölzle's talk was mostly about all that Google has built in the years since then, the implicit message was that if it had been born into a tech ecosystem capable of meeting its needs, Google wouldn't have spent years in survival mode. It wouldn't have had to invest so much money in infrastructure engineering so it could build so much foundational technology itself.
Today, such an ecosystem arguably exists. Companies beginning life today -- and even those looking to rejuvenate themselves after decades in existence -- can now do the things they want to do technologically without feeling like that pre-2003 Google, constrained by concerns over bandwidth and money.
This is largely because Google and its hyperscale peers have done the legwork and created many important technologies -- especially in the world of data management -- that are now available as open source projects. Things like Hadoop, Cassandra, Kafka and Docker. Often, they're also backed by commercial software and smart vendors.
At Mesosphere, we use an open source technology -- Apache Mesos -- to create a Datacenter Operating System that makes is easy to deploy, manage and scale these types of systems on a single cluster of machines. If Hadoop made it so you don't have to worry about too much data killing your IT budget, we make it so that you can give your actual business the attention it deserves because you don't need to worry about a dead server, or an influx of users, bringing down an entire service.
If you're so inclined, you can use the Mesosphere APIs to build out a specialized database or processing engine a la Google, leveraging the inherent scalability and fault tolerance of Mesos to ensure it's reliable.
The beauty of all this is that much of the hard work has already been done. Characterize Google, Facebook and their peers how you will, but there's no denying that Mesosphere is an infrastructure company. And we are one so that you don't have to be. Unless, of course, you really want to be.

Ready to get started?