Cloud Business: Must have

Amazon announced AWS earnings in April for the first time ever since it incepted that line of business 9 years ago. Quarterly net sales rose by 49% to $1.57 billion and operating margins were 16.9%, higher than anticipated. There is no reason to doubt revenues will continue to rise sharply in the future due to the immense market opportunity in the cloud services. 17% margins are very healthy considering how aggressively Amazon reinvests in growth potential such as global Data Center footprint rollout.

Not bad for a book company eh?

What I’d like to call out here is gone are the days analysts would call for small single digit margins in cloud. The 17% cited by AWS is also conservative due to their asset expansion strategy, so the reality is significantly north of 17% margins. What this means is Data Centers and MSP’s that have been hesitant to embrace cloud due to profitability concerns now should realize those concerns are no longer valid.

Time to get in the game.

So what does one look for in a cloud platform deployment to maximize profitability and minimize risk? Given the market opportunity, how does one go about full deployment? You can pay IBM, VMware, Cisco, Accenture etc to do the deployment, but then your contribution is heavy handed to their margins and why wouldn’t one try to retain that? Second option is to take on OpenStack development in house or via out sourced model. Either method is fairly labor intensive since OpenStack needs significant development to become commercial grade. Third is to take advantage of the new Cloud Management Platform Providers that have developed very agile and nimble solutions given lessons learned from the decade old players.

Obviously many factors to be considered, but I’d like to suggest three areas:

First is a Hybrid architecture that scales. Architectures that allow access and control of multiple hypervisor and cloud types will become table stakes in the coming years regardless of initial deployments being public or private. Think years out and scaling across dozens of data centers/regions and the ability to pursue the most cost effective cloud pathways dynamically.

Second is Velocity. A cloud management platform worth it’s salt should be able to be fully deployed within 30 days. Caution here is to be sure the commitment is fully functional cloud threaded to all applications. Many cloud platforms claim rapid deployment times, only to initially produce an idle cloud with little to no applications fully meshed.

Third is Cost. Ideally the pricing should be based on a consumption model versus heavy effort/pricing up front.

Pay as you play. It's only fair.