I just read an interesting article that suggests Zynga may made an mistake when they moved 80% of their games from the Amazon cloud onto their own private cloud. The idea was that the Amazon cloud was a great place to start running new games, but if you want real efficiencies, you need a level of platform optimization that is just not possible on a public cloud.
The author correctly points out that Zynga created a new set of challenges when they moved to their own private cloud, but that is more likely a problem in execution, not strategy. Any time you build your own infrastructure you have to anticipate that both the number of employees and their expertise will change. You also have to make sure that management does not get distracted from their main business – creating great games – when they transform another part of their business.
I have always been a proponent of both public and private clouds. Many in cloud computing are very religious in their attitude, singing the praises of their favorite cloud while decrying the shortcomings of the other. Their binary approach is little more than a set of blinkers which prevent them from seeing the market realities of customer needs and complex technologies.
The theory of long-tail marketing suggests that while a few products may make up the bulk of your sales, the remaining small volume sales can add up to a big opportunity. Rather than a classic bell-shaped “normal” distribution, many product sales cycles peak early and then trail off into the distance – hence the “long tail”.
Interestingly enough, it was Amazon that leveraged long-tail marketing with their on-line bookstore. While they were able to fulfill millions of orders for the latest Dan Brown adventure best seller, they also fulfilled single orders of incredibly obscure books. No bricks-and-mortar book store could possibly keep all those titles in stock, but Amazon leveraged their vast warehouses and automated access to publishers to deliver whatever book you wanted. Those onsies and twosies added up to a lot of sales.
Many apps, especially games, have a similar lifecycle. Sales start slowly, become very successful for some time, then slowly trail off until the maker decides to kill it. It seems there is always some hard-core cadre of players who will never give up on their favourite game as long as it is available. Even ERP apps have a similar lifecycle.
Hybrid cloud computing is a perfect match for this sort of lifecycle. You can develop and launch apps on the public cloud, taking advantage of the cloud’s flexibility and low cost. If your app is a success, you can move it to your own private cloud where you can design a highly-efficient custom platform to minimize your costs. As the app’s popularity dies off, you can move it back to the public cloud and re-allocate your private cloud capacity to the next popular app.
The flexibility of cloud computing can help you evolve your platform to match your product over the product lifecycle. Public versus private cloud is not a one-way street. Public, private, and hybrid clouds are not discrete; they can be mixed and matched, changing over time, to deliver the best solution at any given time for each app.
I admit that that is a lot easier said than done, but getting it right can make your company more efficeint, more flexible, and more profitable.
Wally Kowal is the president and founder of Canadian Cloud Computing