While you probably haven’t heard the term very often, “cloud-in-a-box” (CIB) is one that may get your attention going forward as companies look to deploy private (and possibly public) clouds more quickly, efficiently and with as little administrative overhead and cost as possible.
The CIB idea’s been around for years now, but ever-more compact storage devices (solid state drives (SSD) in particular) and multi-core processors are really enabling the development and deployment of these small but hyper-powerful systems. The idea is pretty simple: put everything you need to run a cloud into a small box (2U in some cases) and plug it in. Do a couple of connections to your network and some set up and, presto, just like that you have a fully functioning cloud.
Scalability is another big plus for CIB. As more capacity is needed, you simply plug in another unit or add more storage and off you go using the same management software and network connections. This is very attractive not only to IT managers and administrators but it opens up the door for a lot more automation of workloads, releasing actual humans for more value-add tasks like developing new applications or optimizing the network to take advantage of all this new-found horsepower.
To date, deploying a cloud means IT managers must cobble together the servers, switches, controllers, hypervisors (virtualization), security, fibre channel and Ethernet, racks and storage, and management software — all from different vendors and then make it all run together. This works fine for the most part but it doesn’t scale very well. Say you want to spin up 50 or 100 virtual machines (VMs) to test out a new application before going live, that takes a lot of time and resources that must be procured through, and provisioned by, IT.
If you have a CIB that is already pre-configured to run as a unit, then it is much easier to deploy the needed VMs for a short period of time and then put the resources to use on another project. Or, as many companies are doing, you could just dedicate a CIB to your developers and set them loose. No need to put actual humans onto the problem every time some project or other needs compute time.
One of the other big attractions of CIB is its on-board storage. Data is exploding and storage area networks (SANs) are expensive to deploy and manage. CIB integrates storage right next to the CPUs that need it. This not only saves a ton of money by not having to provision a new SAN or adding to your existing one, it also means that read/write speeds (known as IOPS) are increased since the data only has to travel a few centimeters to get from where it is stored to where it is needed.
CIB is the logical extension of the converged infrastructure (CI) trend you may be hearing about where IT is leaning more and more on pre-configured racks of optimized best-of-breed hardware and software that come ready to go out of the box. Now, while this is similar to the traditional approach of deploying a cloud, the big difference is these units are already factory-optimized to run in your environment. This saves your IT teams a lot of time and effort.
CIB comes in many flavors from full-on data center racks built out with 10G switches that plug into your data center’s existing fabric to 2U rack mountable units with 80 CPU cores and terabytes of SSD storage that can be used as stand-alone power houses. Because of this, productivity goes up, resource utilization goes up, and go-live times are shortened in some cases from months to hours (really). It’s just a matter of figuring out what you need and putting in the order.
