The public cloud is so big, even Oracle’s Larry Ellison has finally given up on fighting it and now hopes to dominate it. But, not everyone thinks it’s going to be all public cloud, all the time.
Cloud Foundry, for example, has made a nice business helping enterprises build private clouds. I therefore challenged Cloud Foundry Foundation CEO Sam Ramji to make the case for private clouds. Full disclosure, Sam and I are friends. We met more than a decade ago when he ran open source strategy at Microsoft reporting to CEO Steve Ballmer, the man who infamously described open source software as “a cancer.” That’s a tough job, but Sam’s a tough executive.
Perhaps that shows in his interest in leading out on private clouds. Two years ago, he took up the CEO job leading the Cloud Foundry Foundation. Cloud Foundry is an open source cloud computing Platform-as-a-Service (PaaS) originally developed at VMware before being spun out by Pivotal as a non-profit backed by many leading technology vendors, as well as Global 2000 enterprises, such as Allstate, Ford, GE, Home Depot, VW, and more. That’s great backing, but is it simply a recognition that enterprises find change hard and want to cling to their private data centers?
Why go private
TechRepublic: Why wouldn’t everyone want to ride the downward cloud pricing trend versus the risk of building and running your own private cloud(s)?
Ramji: Multi-cloud is real. Large companies, those in regulated industries, and companies who compete with Amazon all have good reasons to use private clouds in combination with public clouds.
Large companies with sufficient skill can now beat AWS pricing through internal operations. It takes investment, but thanks to technologies like OpenStack, Cloud Foundry, Apache Mesos, Docker, and Kubernetes, the core technology barrier to running your own cloud has collapsed. Once upon a time, only Amazon really knew how to do this. But, like all technology innovations, the knowledge has diffused throughout the industry. Amazon is no longer the king of low prices for utility computing. And with that shift, private cloud is a smart option for specific workloads.
Regulated industries such as financial services have strong motivation to stay away from the public cloud. A good example here is BNY Mellon, a global bank that provides services for $28.6 trillion in annual trades and settlements. They need to provide a globally secure, audited, regulated banking platform as they serve thousands of large companies around the world. As their CIO Suresh Kumar told me, “There’s no way we can trust all of this to a public cloud.”
Companies who compete with Amazon and run their digital business on AWS are taking substantial risk. [Amazon CEO] Jeff Bezos has made it clear that “your margin is our opportunity” and his company gets real-time indicators of other companies’ activities on AWS. Network traffic load, compute intensity, and growth of storage correlate with business progress. Recently, Amazon starting to compete with a longtime customer, Maersk–one of the largest shipping companies in the world–by starting Amazon’s Shipping unit. Does it make sense for Maersk to continue to pay their competitor to capture their business data?
A question of ‘and,’ not ‘or’
TechRepublic: You represent an alliance of cloud vendors and Fortune 500 consumers who champion a PaaS technology that most people only use on private clouds. Are your members embracing Cloud Foundry as a security blanket for their legacy enterprise computing infrastructure while their dev teams quietly put jobs in the public cloud anyways?
Ramji: Cloud Foundry is deployed on public and private clouds.
Amazon, Microsoft, and Google have all contributed code to Cloud Foundry to make sure it runs well on their clouds. Pivotal, the company which championed Cloud Foundry and remains a major contributor, runs Pivotal Web Services (“PWS”) on Amazon. The Home Depot runs Cloud Foundry on Google. Ford runs Cloud Foundry on Azure.
Enterprises have many workloads. Some workloads run best on one specific public cloud or another. For example, we’ve found that Google has far and away the best internal network performance. So, a network-intensive workload should probably run there. Microsoft has GPUs available in Azure for video rendering and HPC workloads. Amazon does a very good job at storing files and objects and distributing them globally for fast access.
Enterprises also have troves of legacy data and applications. The reality of IT is that 90% or more of the budget goes to maintenance. That leaves very little for new development. Enterprises choose between migrating legacy applications to the cloud and writing new applications. For a new app that makes heavy use of legacy apps and data, it is architecturally brilliant to locate your cloud platform on-premises, right next to the old, trusted, legacy servers. That doesn’t prevent you from also having Cloud Foundry run in public clouds. Per our recent survey, most enterprises using Cloud Foundry run it in multiple places simultaneously.
Making sense of multi-cloud
TechRepublic: What do you mean when you talk about multi-cloud? Why would enterprises want to bother with managing their computing workloads across different clouds?
Ramji: Gartner Research found that the average enterprise currently uses 4.6 clouds. Multi-cloud is not a concept, it’s an active reality. There are three main reasons that I see for this.
First, enterprises are, by definition, large companies with tens of thousands of employees and multiple divisions in multiple countries. Despite many having established a “no cloud” policy years ago, here we are–the developers have already gone out and done it, and they’ve picked clouds as they’ve seen fit. It’s only now that the bills are getting large that CIOs are getting clear on just how widespread, and relatively unmanaged, this usage has become.
This leads to the second point: CIOs are becoming aware of the size and breadth of cloud spending and they must get that under control. Given their experience of being under duress from monopoly vendors in the past (such as Oracle and Microsoft), they don’t want to give in to a new one. Multi-cloud is a vendor management strategy as well as an architectural pattern.
Third, it’s an architectural pattern that has evolved because different public and private clouds are good at different things. Enterprise IT’s job is not to take a simplistic approach for the sake of saving themselves some hassle. Enterprise IT stands to make the business as effective as possible by exploiting technologies that optimally support the business–and then taking on the responsibility of managing the resulting complexity. Enterprise IT has never been an easy job and no one in the business thinks it ever will be.