The intriguing Dell PowerEdge VRTX “Modular Infrastructure Platform” device can be a bit of an enigma for IT departments. Just understanding exactly what it is can be difficult. So we will start there.
The VRTX is a 5U rackmount or optional pedestal blade chassis that can hold up to four blade servers along with a shared DAS storage array that connects to all of the blades. What makes the VRTX unique is that it mixes a traditional SAS-based DAS drive enclosure (equivalent to a normal 2U drive array) into the same enclosure as four blade servers. Typical blade enclosures focus on a large number of blades in a single enclosure and either attempt to fit storage into the blade form factor or expect that storage will be handled externally.
The VRTX tackles traditional blade problems by reducing the number of blades to a small enough number that the blade lock in problem is mostly eliminated, while addressing the traditional shortcomings in providing storage for those blades. This makes the VRTX a unique and interesting offering. It is a surprising amount of power and storage for a small, single chassis design – with a dozen large format, or twenty five small format hard drive options!
The VRTX can hold up to four M630 dual processor server blades, or two full height M830 quad processor server blades. In either configuration, the VRTX can hold up to eight total processors.
It is important to note that while the VRTX has four blades in the chassis, and has an option for redundant RAID controllers (PERC8) there is still a single, shared storage device in the back and a single shared chassis. So while the server can be incredibly reliable, especially as the storage is internally cabled SAS connected DAS, it is not a true high availability system on its own. It is often confused as one, and this leads to many of the VRTX discussions that we see. It is, however, a very reliable, high end device and more than adequate for a large number of deployment scenarios.
So while there is no doubt that the VRTX is really neat, super cool and one of the most drooled-over compute devices of recent years, what we really need to know is when do we choose to deploy one?
ROBO (Remote Office / Branch Office) This is probably the number one deployment scenario for the VRTX. A remote “datacenter in a box” with enough processing power and storage to power nearly any corporate branch office. A key benefit here is that systems can be preconfigured centrally and shipped as a single unit, ready to go at a branch office. Branch office IT staff (or simply remote hands) need to do nothing but plug in power and networking to this single box. With powerful remote management tools, central IT staff can easily work on the VRTX and its components remotely while ROBO staff need not be encumbered with complex rack and stacks, cabling, configuration and troubleshooting.
Many remote offices lack the dedicated server space of a central datacenter as well and the VRTX takes this into account. It can be racked, of course, but in its pedestal mode it has casters and can sit easily on its side on the floor, even under a table or a desk. The VRTX is quiet and designed to be deployed to work spaces, unlike traditional rack mount servers.
Decision Clusters Business intelligence, using tools like Hadoop, is rapidly becoming more and more important to businesses of all sizes. Because the computational needs of decision systems can be intense and abnormal compared to other workloads, it can be an ideal workload to split off into a dedicate system that is performance – designed around this use case.
The VRTX’s blend of eight CPUs in such a small chassis with a large, high speed storage array is ideal for many BI type systems, especially for medium sized companies. BI does not benefit from high availability infrastructure and so is often unnecessarily costly to run on a more general purpose platform.
VDI Much like how decision systems often benefit from having their own infrastructure that is balanced different from general computing, VDI (virtual desktop infrastructure) often benefits in the same manner and, again, the VRTX is very well balanced for this workload where storage is often heavily shared and front end workload is generally load balanced at an application level.
VDI generally needs special hypervisor considerations, GPU considerations, heavy CPU and RAM resources and shares disk space in unique ways that fit right into the VRTX’s wheelhouse. A VRTX can be a great addendum to a large general purpose computing infrastructure when VDI is needed to be added.
SMB In the small business space the VRTX is a less common tool as it often brings more power with it, even in a small installation, than an SMB can often consume. But for SMBs that are seeking a mix of heavy CPU and RAM resources, especially in a small package, or those that are looking for a physically convenient single box that does not need to be stored in a dedicated server room, cabinet or space the VRTX can be a very good fit.
SMBs rarely need high availability solutions and the VRTX’s “higher than normal” availability options can fit the bill nicely for workloads when needed and high computational power for more workloads, when it is not. The VRTX can handle very large databases or a very large number of disparate workloads. For SMBs needing that extra horsepower, or to maintain testing environments, the VRTX can work well.
Guest Post: Scott Allan Miller