Recently I’ve fielded quite a few inquires on my thoughts on hybrid and public clouds. Not only am I sold on the hype and promise that it can deliver on, but I am equally enthusiastic about how it can provide an alternative platform for delivering IT, the elasticity it offers, and how it can address hosting workloads that require scaling. On-demand infinite resources, any time - under a pay as you go cost model – is wickedly awesome. Wickedly awesome – and sometimes wickedly expensive.
When customers ask me about hybrid clouds, I typically turn the discussion around and ask them a series of questions: What type of workloads would you move into a hybrid or public cloud? What type of elasticity do you need? How much excess capacity does your current data center have? Do you have security or regulatory concerns?
The responses I receive vary from “We need seasonal-variable elasticity with additional capacity in the range of 30% to 50%”; “Security and regulatory concerns will prevent us from leveraging shared services”; to a more humorous response of, “Our CIO thinks it would be really cool to offer this”.
Then to the million-dollar question – with the widest range of responses - “What is your current cost model for delivering Virtual Machines/IT services - Do you have a chargeback model, and at what speed and agility are you able to deliver them at?”
Why are you asking? The reason I am asking is because if speed, agility, elasticity, and costs are drivers behind cloud adoption, shouldn’t you understand internal costs before considering external clouds? If you want to compare provider A against B, and B with C, shouldn’t internal ‘IT’ also be compared to A, B, and C? I find it intriguing to often hear, “We really don’t have an internal cost model for VMs”, or “We don’t measure the time it takes to deliver a VM and it really depends.” Depends on what? Depends on who fulfills the request? Depends on the lunar cycle? Or depends on a whole bunch of variables you don’t understand? I apologize for being critical, but shouldn’t IT understand what costs are associated with delivering a service/VM, and how long on average it takes to deliver it -- before offloading this task and cost to an external provider? In my opinion, SLAs (including performance and availability) and costs should be at the core of making any informed, outsourced ‘cloud’ decision.
With infinite resources available through cloud providers, I always like to caution consumers that unexpected costs are not only a risk of cloud computing, but also a dark reality. With nifty incentive plans to deliver elastic-computing resources to unassuming consumers comes the dangerous and costly side of cloud computing – scalable, infinite costs. Everyone is being rewarded to deliver - I mean bill you - for readily available, fast, elastic and scalable compute resources. At least internally there is something called “available” capacity and usually some type of procurement process for adding additional capacity.
Before buying into an external cost model that often requires 5 lawyers and 3 accountants to understand, I encourage everyone to evaluate internal delivery models with costs, speed, agility and internal elasticity in mind. In doing so, you will have the necessary data to become a good steward of IT resources with the ability make informed decisions on when and how you can leverage external clouds, and how to more effectively manage and optimize internal resources and processes.
Understanding costs is at the core of curbing consumption. It continues to work well at the Appstore, supermarkets, gas pumps, and neighborhood Fourbucks. Isn’t it time we leverage its benefits to curb and more effectively manage the consumption of IT resources?
