How to move the Open Compute architecture to an Enterprise level 2

How to move the Open Compute architecture to an Enterprise level 2

I start the second of those posts that I committed to write around these new solutions, OCP or Hyper Scale. The truth is that they have been a few months since very interesting in which I believe that every time is taking shape and more background.

After the first approach to these new data centers 3 were my main conclusions:

  • Capex reduction and Opex optimization. Analysis from the point of view of the Facility brings benefits both in initial investment and in subsequent operating expenses, including energy.
  • Uptime and Open Compute. The official answer of Uptime Institute is that these new architectures can be certified without problems. The unofficial is that the Uptime's requirements to certify them, throw down all the benefits of the new architectures. I am going to propose not to re-influence much in this matter.
  • Legacy and IT. It was logical after the first approach from a Facility point of view. What about the IT part? And of course, what about my legacy ?, which still exists and I think for some time.

These main conclusions leave me a good taste. Think that as I said in my previous post, for the first time and from a Facility approach, it is the end users themselves who need to join the first Facility analysis to IT and analyze the benefits from a holistic point of view. What Social Networks Do not Join :)

On the other hand, the most negative, although expected. The entities that are in charge of certifying / standardizing will be the ones that take the longest to give shape to this, logical, in the face of new situations they have to adapt ...

Work from now begins really, the product exists, but now it is about moving the industry, preparing integrators and above all fitting the companies' IT architecture to a new infrastructure completely adapted to all those new concepts that go beyond that first Cloud based server virtualization .... DevOps, containers .... If you think about it, it's funny, although long ago I read a publication where it was recommended to stop talking about Business in the IT world, it turns out that in our case (Schneider Electric) watts, we hooked the bits, but now the code is also interested, something good has to be there.

And that is the exercise that I propose for the next post. Evolve the experience upwards, without still talking about Business, which obviously will end there, but we will see the benefits at each of the above levels.

A major event has happened in recent months and is the official launch of Open19, the hyper scale version of Linkedin. Conceptually the electrical architecture is very similar to Open Compute, since as in this is moving the electrical safety to power shelves in rack with batteries, but the difference is that Linkedin allows us to reuse the traditional 19-inch racks and for this mounts a chassis in the racks where they host the IT equipment that as in Open Compute do not have power supplies. Is time to evaluate the advantages and disadvantages of each of the options at the rack level, but what does not change are the main and important changes, the rack is becoming a key element to incorporate the power and electrical safety of IT devices. I leave you some images of Linkedin's own engineers.

Finally, I'm going to end up teaching you that these new architectures are already changing the current data centers. The following image shows HyperPod, a self-contained solution from Schneider Electric designed for the Internet Giants but now ready for the Enterprise Data Centers. There are two major trends currently in the deployment of IT solutions, the first is to define standard PODs at the design phase that let the data center growth with predictably and according to the needs of each moment and the second is the rack-based deployment (so do the Hyper Scale data centers, but it is also the philosophical one of the hyper-convergence solutions), i.e., we do not deploy the rack, install the IT equipment and make the aisle containment, the idea is that the integrator will gradually incorporate services that prepare the racks to deploy them in the customer's home already configured, just connect and boot. HyperPod is designed to be able to have everything ready for when these racks arrive, so that containment, electrical distribution, network, air or water piping, etc ... is prepared and only needs to locate the rack with the IT equipment and connect.


To view or add a comment, sign in

More articles by Jorge Jimenez

Others also viewed

Explore content categories