Five Easy Steps for Implementing the Cloud

Anyone recognize this scenario?  A CEO goes golfing with other CEOs and she comes back and she says, “I want to go 100 percent cloud.”  We technology folks joke about this, because it’s both so stereotypical and it’s also so true.

Frankly, I’m tired of all these different responses and opinions to “what does Cloud mean to you,” because I cannot help but continue to refer back to the National Institute of Standards and Technology definition of cloud.  Besides the idea that it sounds so official, it’s also really the best answer.  For those that don’t know, I don’t blame you, but it would be wise to pay attention and take a look at an oversimplification of the five essential characteristics and implement them.

• One-demand self-service
  • The ability to provision servers and storage without human intervention
• Broad network access
  • Resource can be accessed by many different devices, such as phones, tablets, laptops, desktops, and servers
• Resource pooling
  • Servers and storage are shared across physical hardware without location limits
• Rapid elasticity
  • The ability to grow and shrink, even in an automated fashion
• Measured service
  • Reporting and monitoring of the usage of each component

If the above sounds a little daunting to implement, most of it is not.  It can be done using private cloud resource, public or a mix of the two (hybrid).  Below are five easy steps to get there. If you want to get to 100% cloud, here’s how:

• Create a self-service portal using existing tools today. That’s it, folks, don’t make this too complicated.  It’s just a website where you can create a server, based on your requirements, and when you hit submit, the server is created.  You don’t have to give this out to everyone in the company either.  You can keep it inside of IT only.
• Build the largest pool you can of server, storage and networking resources. Those companies aiming for 100 percent virtualization totally get it.  Having one big pool does not mean “all your eggs in one basket.”  It just means that you have the ability to move anything around to any location inside your data center.  Unless you have over 1,000 VMs, this is not hard to do.  Just make all VLANs accessible to all servers and zone all your storage to all your servers.  Really, that’s the crux of it.
• If you build a big pool, you’ll naturally be elastic. If the business says, “We just acquired a company and need to build 30 servers – what’s the cost, what’s the impact and how soon can we have it?”  If you design it right, the answer is: “no impact, you can have it today and it costs “X”.  Infamous bucket of water analogy:  Take a five gallon bucket, and use it to pull water out of a bathtub.  There’s a noticeable drop in the water level.  That same bucket of water removed from a swimming pool will have a trivial effect of the water level.  You can easily add or subtract many buckets of water from a large pool.  A larger pool is more cost effective, efficient, etc.  And I recommend cost calculations based on the entire pool over three years.  In other words, if your 20,000 gallon swimming pool costs $20,000 to build and $3,000/yr to maintain it, add up all the costs for the entire environment over three years, ($29,000) and then just factor the five gallon cost of $1.45 per gallon to be $7.25/bucket.
• Measured service is just a fancy way of saying “monitoring and reporting,” although we want to make sure we can get granular on the details. If you want to know, for example, the costs of running a service for a shorter timeframe, or want to start billing using peak times, you still use the average numbers, but you add another factor for startup/shutdown costs, and you can also add a factor for usage.  Again, there are lot of tools out there today that have this feature set and from many vendors.

And there you have it – a guide to cloud, based on a real good definition.  Notice, there’s no requirement in here to go to a public cloud.  Public clouds do arguably have much bigger pools to work with, and they can be more cost effective, but not always.  And let’s be honest, most folks don’t have a requirement for extreme elasticity where the environment is doubling or tripling in size overnight and then shrinking back down the week after.   And there’s no mention of products or other marketing terms like software defined storage, virtual networking, converged, hyper-converged, etc.  But it is the heart of cloud, and unless of course you get push-back from people telling you that their definition of cloud is better than the National Institute of Standards and Technology, this is a good place to start.