The ScaleIO Story

Over the years, the team of ScaleIO (pronounced “Scale-Eye-Oh”) founders has been involved in various situations in which a customer had acquired expensive SAN products that were standing unopened, still in their cartons and shrink wrap, in the data center. The reason? Not enough power, cooling, or switch ports to connect the storage. At the same time, thousands of internal disks, supporting just a few gigabytes of operating system each, were spinning in the application servers, hogging power, cooling and space.

What if, thought the founders, we could aggregate all that free capacity on those disks and turn it into a high performance SAN? Out of thin air we could create a SAN that does not require any additional hardware, power, cooling, space, ports, host bus adapter cards or dedicated storage resources. Moreover, the storage would become as inconspicuous and easy to manage as CPUs and RAM. Customers would benefit from reduced complexity and significantly lower costs, and capacity management would no longer be such a pain point. The data center would be streamlined, consisting of servers only, from wall to wall. Defining in advance what hardware will handle compute and what hardware will serve as storage would be a thing of the past, since all hardware can do either or both tasks. Also, with the elimination of disk price arbitrage, customers would enjoy a huge price advantage: disks that run in external storage arrays are 10 to 40 times more expensive than the very same disks when delivered in a commodity application server!

Market developments that promote convergence, like Oracle Exadata, EMC Vblock, NetApp FlexPod, and HP Converged Infrastructure solutions, support the concept of mixing and matching compute and storage—but on a marketing level only. ScaleIO makes this vision a reality, today.

Original ScaleIO Design Objectives:

  • Performance (we have achieved this goal through parallel architecture)
  • Scalability (essential for modern data centers with thousands of servers)
  • Elasticity (the ability to add, move, and remove servers on the fly, an extremely important operational advantage)
  • Brownfield support (the ability to support servers, disks, networks, and other components at any point in time—what the customer has in the data center today and what is added to the data center in the future)
  • Elimination of storage vendors' hardware lock-in
  • Intuitive data center operations (any IT administrator should be able to manage the entire data center stack, from applications to storage—and we have achieved this goal by turning all tiers into software applications running on a server)
  • Elimination of the pain of capacity planning (a major issue with traditional storage, but with ScaleIO, planning capacity is as easy as planning RAM and CPUs)
  • Simplicity (a rapid installation, which any Linux administrator can manage, and the ability to work on any hardware that the customer chooses)