Technically Speaking

 So this week is storming right by and is capped off with a trip to Fort Collins CO to learn about 3Par and Lefthand and how HP does storage.  I am interested to learn more about both of these products, and even storage in general. There will be lots of useful discussion for sure which is the best part.  

Bringing bloggers together to discuss and learn about HP products and the tech that makes them work is great and I’m glad to participate.  Like others attending I have read some of the white papers covering both 3par and Lefthand but have no experience with them. No worries, I’m sure it is coming soon. After going to CO, next up is Tech on Tap I am very much looking forward to the first event and hope it is the first of many.

Time to pack and charge my iDevices.

While spending time in Houston at the #HPProliantDay I found the discussions around power and cooling to be most interesting (until they started in on storage at least).  Having recently purchased some Proliant servers, I was quick to discover the plastic barrier to visibility that lives under the lid of the box.

I was quite confused by this plastic as it seemed to get in the way and just cover up the parts.  No where on this random item was there a label explaining what it was for.  Turns out they had a plan… controlling air flow within the case to allow better cooling performance and reduced power usage for cooling.  Makes a lot of sense when you find out the actual reason for something.

Then there was the discussion of power, which involved the intelligent power distribution unit (PDU).  These devices monitor the power sent to G6 and G7 servers and will also indicate where the server is plugged in, which rack it is in based on PDU, and a few other features.

Why care where a server is plugged in?

Suppose you have a server with redundant power supplies and you plug them both into the same power strip or outlet group, when that device loses power the server will go offline.  With these PDUs, the devices themselves are managed through a browser application and will report an error if both of a server’s power supplies are plugged into the same PDU.  This helps ensure redundancy for the server by removing the single point of failure at the power supply, or outlet level.

You can also determine how much power a server is actually using and create a cap for certain outlets ensuring that these outlets never exceed a certain amount of power, which will not only cut down the amount of overall power used, but reduce the likelihood of blowing breakers because the power sent will not exceed the cap on the outlet.

Being interested in storage I never gave much thought to power and cooling, until I unboxed a SAN and some new servers and plugged them in.  Shortly after doing so, I had blown a breaker and was immediately informed by co-workers that something was wrong with everything.  Sure this was something I could have avoided and have since fixed, but being able to get a look at power consumed would allow me to make better use of the total amount of power provided to the server room.  If the power being sent to the room, is a certain amount, making the best use of that amount of power is a great idea, but reducing the power drawn by the devices in the server room to save money on power usage of the organization is even better.

I am looking forward to investigating the intelligent PDU in the future to see if it might provide the almost immediate ROI I am hoping.  There will be some spending up front to implement these items, but over time (and quickly I am sure) I should see enough savings in power to pay for the devices (and add them as needed).

The Quick Specs for the intelligent PDU are here

Today we went to the new Dallas Cowboys facility to learn about how HP and the Cowboys built a very growth ready infrastructure to help the Cowboys achieve their needs for today and many years to come.

Converged Infrastructure is the way HP brought it out. The concept is great, the theory behind this idea of converged infrastructure is that components are modularized to allow pieces to be plugged in and out as needed.

The concept relies on several key components including:

Storage
Servers
Network
Power and Cooling
Management

These are the key pillars of the concept. The implementation relies on virtualization and vendor openness to ensure that the concept can be achieved if all of the existing gear in a site cannot be torn out and replaced. This doesn’t have to be a rip and replace all HP world so existing shops can work with some of the technology that they have in place today to achieve this concept.

Leveraging technologies like virtualization allows the environment to run more reliably on a smaller server footprint which can reduce power consumption costs dramatically.

The Dallas Cowboys organization has three data centers to ensure top level disaster recovery is possible and to maintain reliability across all of the businesses they support. No, the Cowboys IT staff do not just support the football team. The Jones family owns over 30 businesses in total and has 96 locations all of which move their data back to these data centers.

Where I see problems in getting there

The converged approach is a great concept, but how does one get started? The rip and replace method seems the best way, even though converged infrastructure aims to bring what exists already and the new additions to your environment you may be considering.

I suppose the replacement process happens over time and gear is replaced as it exceeds it’s useful life, avoiding rip and replace.

Because the Dallas Cowboys were changing facilities and have a considerably larger budget than many organizations, the all at once method worked well and provides a fantastic case study for HP and a kick ass datacenter.

I would recommend touring the stadium if you get chance, even if the datacenter isn’t on the public tour, the football part is amazing. One thing I am curious about is the amount of bandwidth they have. In the private cloud they use, I would guess it’s better than a reasonable connection.

Images from the tour are coming soon.