In January I attended a webinar by Forrester analysts James Staten and Sean Owens on the economics of cloud computing. The slides and recording of the presentation are linked here and here. Although the research was commissioned by Microsoft and thus had an Azure flavor, many of the insights are equally applicable to other cloud platforms such as Amazon or Rackspace. What was unique about this presentation was that it was targeted at economic decision makers vs. technology leaders. The presentation is well done and is insightful.
The first part of the presentation has a good basic introduction to the economics of cloud computing. With a public cloud you only pay for what you use, cloud platforms allow for development teams to self-provision, and perhaps most importantly, a public cloud allows an organization to instantly add capacity should demand materialize.
Although I am a huge proponent of public clouds there are pitfalls. Some of the economic reasons NOT to use a public cloud include:
- Ease of running up a giant bill. There are horror stories of teams running load tests that accidentally leave a test server enabled and incur what Forrester referred to as the “shock bill.” An, active, yet idle VM costs the same as one that is being used.
- Not using existing investment. The ease of using a public cloud makes it easy for local teams to “roll their” own in lieu of using existing infrastructure that has already been provided for them by their IT group.
- Loss of control. Public clouds are just that – public, and may not have the same level of security found in a hosted solution. While the cloud provider may ensure that the operating system is fully patched they cannot guard against weak administrative passwords.
- OpEx vs. CapEx. Public clouds are almost always considered an operating expense where an on-premise solution is typically capitalized.
These issues should NOT be used as excuses to avoid public clouds. Indeed there are “patterns” and best practices to ensure that the organization is making the best possible choices.
The second part of the presentation was insightful in that it culled out some of the indirect benefits or perhaps less obvious reasons to use a public cloud. For example,
- Cloud computing is cool. Developers want to work on cutting edge technologies and availing them access to Azure / EC2 helps with recruiting and retention.
- Public clouds push local IT groups. Some organizations struggle with non-responsive / bureaucratic IT groups. The ability to “outsource” hosting (for public or staging servers) puts pressure on IT departments to provide better customer service.
- Geographic distribution. Although not discussed extensively this presentation, services like Azure and EC2 allow organizations to host in a geographically distributed way. This capability, particularly when combined with a caching strategy, is useful for improved performance for global customers and ensuring that an organization is compliant with local privacy regulations.
Forrester is a very careful organization and the economic claims in the presentation are no doubt true. On the other hand some organizations may find that their “mileage will vary.” Cloud computing is nearly always more cost effective than buying new physical servers. A hybrid public/private hosting strategy, levering platforms such as VMWare and Microsoft’s own Hyper-V virtualization system, could be more cost effective for established entities with an existing investment in on-premise hosting. Private cloud platforms support the convenience of self-service and dynamic growth but have minimal operational costs. (There is a fixed, typically capitalized, cost to host the server.) This type of solution is very appropriate for development and QA groups that need non-production servers.
Bottom-line for a fast growing start-up there is nothing better than having no hosting footprint yet being able to scale on demand. For established entities the public cloud also makes sense but should be considered in the context of existing investments and used where it makes sense.