I made a presentation at the recent Storage Visions conference which gave me an opportunity to think about how rapidly Software-Defined Servers are evolving and how perfectly they fit with software defined storage architectures.
In the earliest versions of TidalScale’s inverse hypervisor technology, our primary focus was on virtualizing CPUs and memory – the essential server components that until recently had been absent from Software-Defined Data Centers. Since servers were the missing piece of that software-defined puzzle, we zeroed in on finding a way to make it easy for you to right-size your servers to fit any workload or data set. By allowing you to combine the resources of multiple commodity servers into one or more Software-Defined Servers, we turned what once was a fixed resource into one that can be combined and configured on the fly, and all without a single modification to either your operating system or your application.
Part of that picture was virtualizing the direct-attached storage associated with the servers you were combining. This was useful because, say, in a Software-Defined Server made up of 10 nodes, we would assign a disk installed in one node as the boot volume. Another disk in a different node could serve as the data volume. To the user, this all appeared as one large server, with the associated resources like CPU, memory, storage and network all available to applications as one.
But it quickly became clear that by tying the boot volume and data volumes to certain systems within the virtual server, we were limiting our options for configuring the Software-Defined Server. The resulting system was certainly flexible, but it wasn’t entirely stateless. Some nodes had to be used for certain purposes because of those pesky disks. For instance, you couldn’t float the workload across all servers because they weren't entirely stateless. We knew we could do better.
And with our recent major software release, featuring the TidalScale WaveRunner console, we did. With via the WaveRunner GUI, we give users the ability to integrate their Software-Defined Server directly with a centralized software-defined storage architecture. WaveRunner takes adventage of iSCSI, which presents a boot disk housed in the central storage repository as if it were locally attached. This allows us to bypass server-installed disks entirely, thus ensuring that every Software-Defined Server is built from a collection of truly stateless servers available to handle any workload.
With the WaveRunner console, TidalScale has managed to tie all the software-defined elements of a server together into a single virtual resource. CPU, memory, storage and network now are under the complete control of the user, with a simple point-and-click interface allowing virtually anyone to configure and boot a Software-Defined Server in as little as a minute.
The old configuration limitations are gone, and with it the choke points that could have impacted application performance. For example, when you run multiple tests on Jenkins, the continuous build and test automation tool, the ability to take advantage of the whole of the resources on your Software-Defined Server can have a dramatic effect on your ability to complete multiple runs simultaneously rather than sequentially. If you have multiple jobs to run, and most require a fraction of your available resources, there is now nothing preventing you from utilizing all the resources on your Software-Defined Server pool. We saw this in action with one example, in which we were able to increase the number of tests completed during a weekend from 12 without WaveRunner to 250 with it. That’s an impressive 21X boost. (Download our recent white paper for details.)
Because of this impressive effect on our operational efficiency, TidalScale decided to exhibit at this year's CloudExpo - come visit us at booth #309 in the Santa Clara Convention Center Oct 31-Nov 2.
The TidalScale WaveRunner console turns any data center into a truly software-defined data center. Find out more by viewing this video and learning how virtualization has just taken a major step forward.