Broken Promises

There's an awful lot of discussion going on in the Intertubes about the concept of Enterprise Clouds, but most of it is confusing private (single enterprise) clouds and the concept of a multi-tenant cloud platform that is built to help companies avoid the issues associated with “designing to fail”.  One of the more useful descriptions of the difference between private and “enterprise” clouds (at least that I’ve found) is from Simon Wardley in his “Private vs. Enterprise Clouds” post. It’s worth taking the time to study his thinking, as I believe he has a firm handle on the situation.

As for the public vs. private debate, I just can’t get behind the private cloud solutions.  The economics aren’t there, not to mention the very real challenge of finding enough multi-disciplinary talent to effectively build and maintain private cloud stacks. Fundamentally, I believe that the concept of a private cloud is based on the collision of “Not Invented Here” thinking within enterprise IT departments and the lack of adaptability of the companies providing them their IT infrastructure and service management products.  Two wrongs don’t make a right.  

To me, the main difference between the commodity cloud approach, and the enterprise cloud approach, is in the availability attributes of the utility service itself.  It’s the economic conditions created by the commodity providers themselves that make it very difficult to offer anything other than an environment where tenants are strongly pushed to “design for failures”.  This may change over time, either due to the commodity players deciding that they will take a significant margin hit to get a larger share of enterprise IT spend, or due to technical advancement in the platform fabrics themselves that improve reliability and availability of tenant VMs / data.

Unlike private clouds, I’m a firm believer that enterprise clouds have a role to play in today’s market. A window of opportunity for enterprise clouds exists right now, regardless of the commentary that points out it’s lack of adoption.  I’d say that the lag in (not lack of) adoption is really caused by providers not having (until recently) the clarity of purpose to deliver viable and scalable enterprise cloud platforms.  Now that adoption is starting to pick up, the providers are responding with appropriate investment in this area (with even AWS investing in features that better align it with the needs of the enterprise).  This movement is what is going to give enterprise IT shops the ability to take advantage of many of the “cloud attributes”, while having a solid availability foundation on which to trust their critical systems.

While intended as a point for only believing in the commodity clouds, pointing to CAGR charts for the commodity clouds only serves to support the value of the overall ecosystem of cloud providers.  It doesn’t really make a conclusive argument that commodity clouds are the only model that will win.  Even with the great success of AWS in the market, assumptions that amazon will be in the range of getting annual revenues of $10B US by 2016 just show success in the market and the increasing rate of general growth in IT services. Gartner is estimating that the global quarterly IT spend for 2011 to be $360B, with an average of 4% CAGR.  Given the solid growth of commodity cloud services, and understanding the needs of the enterprise IT buyers, I believe that enterprise cloud adoption will follow the earlier market trends of the commodity clouds for the next several years. There's going to be alot of room in the market for different types of cloud services...

All this is not to say that the market opportunity will exist forever, but the reality is that providing high availability for “legacy” applications in commodity clouds is a significant challenge with today’s features and technology. Just like the mainframes that just won’t go away, the relative usefulness of enterprise clouds will remain with us.  Their important has faded, but mainframes are still critical to many institutions. Enterprise clouds will be very similar, existing as a bridge solution for the lifetime of enterprise “legacy” applications. We're just not anywhere near the top of the growth curve yet.