Dan's Take

Cloud OSes: Benefit or Bane?

Are you gaining flexibility or unwittingly setting yourself up for lock-in?

Several suppliers have announced small, limited OSes designed to live inside a virtual machine (VM). A few examples of this are the recently announced VMware Project Photon, the Piston Cloud Computing CloudOS, the OS component of Red Hat OpenShift and the open source CoreOS.

When I recently had the opportunity to speak with Jim Morrisroe, CEO of Piston Cloud Computing, the company had the following things to say:

"Piston CloudOS is a new type of OS that manages clusters of commodity servers in your datacenter as a single pool of highly available resources. These resources may be shared by any service or framework, including OpenStack, Hadoop, Spark, Kubernetes, Docker Swarm, or Mesos for better resource efficiency, and streamlined operations.

It's clear to me after speaking with Mr. Morrisroe that Piston has its customers' best interest at heart. Unfortunately, I don't always get the same impression when speaking with other suppliers that have recently announced a micro version of the Linux OS.

Although each supplier's strategy and the details of its micro-OS are a bit different, the goals of each supplier appear to be the same:

  • Improve VM performance, making the platform more desirable to customers implementing on- and off-premises cloud environments.
  • Reduce the time it takes to spin up a new cloud workload, which should make its platform more desirable.
  • Improve overall security by reducing the attack surface for Internet-facing applications, which will both benefit customers and reduce the number of embarrassing media stories of security breaches.
Taking Control
What is seldom said is this is an attempt to grab control of the underlying OS and low-level system services from the established suppliers, replacing them with the supplier's own OS. The goals:
  1. Increase levels of customer control.
  2. Provide higher levels of integration, making management and provisioning easier for customers.
  3. Capture service revenue from other suppliers

Although the mix of technology is a bit different for each supplier, this is very similar to the approach Oracle is taking with the Linux OS and Xen hypervisor, and its attempts to control its customers and capture revenues from the likes of Red Hat, SUSE and other Linux distributions.

This move is now possible because enterprise-grade OSes, VM software, clustering software, database software, filesystem software and the like are all available as open source.

But is this necessarily a good thing? As suppliers integrate their stacks of software, could we be moving back to the bad old world of propriety software that makes it difficult to impossible for software to be migrated from one supplier to another, or for workloads to more easily interact with one another? If each vendor has their own OS, file system, networking system, application frameworks, database engines and so on, it can create havoc. What happens when something optimized for a VMware environment can't be moved to a Microsoft environment without being totally rewritten?

I've lived through several variations of this theme that had different "choke points," including variations introduced in "standard" programming languages, database engines, client OSes and server OSes. We'll be back in the world of incompatibility.

Dan's Take: Insist on Open Standards
As suppliers move in the direction of offering their own integrated software stacks, customers must insist on compliance with international and industry standards. If the supplier only offers customers chains (regardless of whether they appear to be made of gold and are decorated with jewels), the customer should recognize them as chains and just walk away.

Many of the suppliers are truly trying to make things better for their customers, and have no intention of building a fortress inside their customer's datacenter to keep the customer penned in and other suppliers out. Customers are advised to make sure this is their actual intention before signing the dotted line and adopting the technology.

About the Author

Daniel Kusnetzky, a reformed software engineer and product manager, founded Kusnetzky Group LLC in 2006. He's literally written the book on virtualization and often comments on cloud computing, mobility and systems software. He has been a business unit manager at a hardware company and head of corporate marketing and strategy at a software company.


Subscribe on YouTube