One of the great things about VMworld is that the greater virtualization and related technologies converge. This includes hardware, evidence by my opportunity to converge on some of the HP events here at VMworld. Today I attended an invitiation-only blogger event on HP’s Performance Optimized Data center (POD). Call it an arms dealer to the cloud at face value, but HP’s spin is that it is an attractive alternative to purchasing new facility space.
Here’s the basic principle, the HP POD is effectively a small datacenter contained in either a 20 or 40 foot long box. This box was initially built into industry standard shipping containers for the generation one (G1) pod, but the generation three (G3) pod is slightly modified from the traditional shipping container to allow for better airflow management, access and accommodate additional length. The 20 foot model contains 10 racks and the 40 foot model contains 22 racks. Speaking in scale, there could be a half-million virtual machines in one of these puppies!
HP’s Wade Vinsen presented to us on the POD. We had an overview of the basic architecture and the sub-components, which are ProLiant SL series servers and StorageWorks 600 Modular Disk Systems (MDS 600). He had a mock-up POD that was a miniature model, which worked well to illustrate the features.
In the spirit of social media, I’ve pulled out my Flip camera and caught Wade in a number of segments. Here is the POD overview video:
POD virtual tour, MDS 600 and right-sizing comments:
One of the questions that I brought up is POD accessibility. I posed questions about swapping out servers and how do you get into the racks. There is 39” from the front of the server to the wall of the POD. From the rear, there is between 16” and 19” to the wall. This means that you have to usually open the door for full access to the rear. The units are capable of being used outside, but I’m not wild about having to check the weather to install a new component or troubleshoot something. The units can be used indoors as well in warehouse settings. Here is a video on the accessibility of the POD:
Bring your money – or not?
The POD is really positioned as an alternative to additional datacenter space. A 20 foot pod starts at around $600,000 empty (no servers and storage) and the 40 foot unit is in the $1,100,000 range. Considering the facility-based alternative, especially as a new purchase; it may not be that far off.
Magic around 50U and power management
One of the measurable that Wade talked about was the magic of 50U that is a design element for the POD. Here is a video on the rack design and the logic around the power management. I had asked why the UPS backups were on top of the rack:
Cost per Megawatt
Wade simply wouldn’t stop with the whole cost-per-Megawatt topic. I don’t deal with the scaling of facilities, so it is new information to me. As you and I know, the business world revolves around ratios and if we can prove a solution that drives a better ratio – we’ll take it. This video is a little dated from the basic principle of the G1 pod, but Wade explains the cost-per-Megawatt in one of the driving factors for a POD to scale out:
Overall on the POD
My questions revolve around accessibility to support the internal contents going forward. This is especially important as partially populated pods are available. This means that the end-user or HP professional services would need to get in there to install systems.
Aside from that critique, I see the POD is an attractive single source solution for quick scaling. Wade floated timeframes in the 6 month range for turnaround from purchase order to power-on. This is the first access to technology like this for me, and aside from the cool factor there is a use case.