The Future of Architecting Infrastructure: Cloud Maps

HP recently hosted a Cloud Tech Day event down at their campus in Houston, Texas, and I had the good fortune of being invited to attend. As a matter of full disclosure, they paid for the entire trip (flight, hotel, and meals) over the course of my 2 day visit. The only documentation I signed was a release to use my striking visage on their videos and repeat my comments regarding their products, but I am under no obligation to be a cheerleader nor was I strapped to a chair and forced to watch marketing videos with NSYNC blasting in the background (Whew!).

With that said, I’ll give a nod to those involved from both HP and Ivy Worldwide for hosting such a solid event to better teach me about the HP CloudSystem product line. I can only imagine the work it takes to corral a dozen people from around the world into a two day window filled with brief time slots.

Do You Like Acronyms?

A whiteboard session covering the acronyms

You had better, because HP absolutely loves them (yes, they even made their name an acronym). I’ve never been assaulted with so many sentences that contained stand alone letters before in my life.

So, HP, please be gentle to my noodle in the future. 🙂

Ok, on to Cloud Maps!

Why Cloud Maps are Awesome

Awesome is definitely the right word for these things. Seriously. The concept is this – you use the HP Insight Orchestration Designer to create an environment in a Visio-esque type layout. Drag physical or virtual servers and storage onto the workspace and connect them using little circles on the sides to create relationships between servers, storage, and networks.

Note: HP usually terms a server as a “service” – which in essence is what a server is doing – instead of a calling them servers or virtual machines.

Each object can have different characteristics, like “only let someone make 4 of this server” or “make sure this network has redundancy”. It’s all checkboxes and fields to enter, very simple stuff. We saw a live demo and a canned demo of the process, and I equate it to be similar to doing a whiteboard design with the enterprise teams, except no one needs to hastily scribble down notes because the end result is a deployable entity. I’m really not doing Cloud Maps enough justice, but suffice to say: you can configure to your hearts content.

The road to building a Cloud Map

Once the Cloud Map is created, the design is saved as an xml file and can be shared to others! In fact, HP also has a repository of some commonly requested maps on their site. This is a great feature because it will nurture a community to grow with known-good configurations that you can deploy. Of course, the underlying hardware will differ from one customer to another, but that can be edited after the map is imported. This will also be a really nice feature for anyone doing consulting as they can create a repository of cloud maps that can be used at customer sites, or as a reference architecture point for their internal teams.

Here’s a picture of what the designer looks like:

Further more, the templates can be made visible from the service catalog to provide user driven provisioning.

Adding More Power

A wise man once said “this needs more power” and then grunted a few times. I think his name was Tim.

I don't think so Tim. Oh wait, yes I do!

As awesome as a Cloud Map is, there is a way to juice it up with more power in the form of Operations Orchestration (OO). Custom workflows empower Insight Orchestration to do even more, like create or delete services, perform silent installations during deployment, send emails at certain points, and create response files. Rather than trying to bake in software into a VM template, you can let OO do the heavy lifting and configuration.

Here’s an example of an Orcale RAC workflow in OO:

A sample workflow
Some of the workflows can get pretty complex - use a modular approach

To me, this is a true future of architecture. End to end configuration of a deployable, automated environment, that can also be granted to the user or manager to provision on demand. I’m very geeked out about this. I imagine grabbing a laptop to meet with the application teams and building one of these, doing a bit of test to ensure it works, and then just becoming the Maytag man for when changes need to occur and being able to focus on the next design.

Our demo session on workflows


I did not write this post to say that HP’s product will specifically be the alpha and the omega, but having this sort of holistic design approach that is scalable and can be shared is definitely something others should be paying attention to in the market. The “down side” to all of this is that it requires a lot of investment in HP converged infrastructure to use, which seems like something you’re either ready to do (hardware refresh, greenfield, really like HP equipment, etc.) or not.

I’ve purposely left out a lot about CloudSystem, BladeSystem and the Matrix Operating Environment (MOE). Understand that in order to harness a lot of this cloud tech (Cloud Maps, OO, etc.) you’ll need a CloudSystem.

Also, if you’d like more on CloudSystem check out this post by Philip Sellers, a fellow blogger at this event, which discusses bursting to cloud.