I’ve written a series of blogs about the economics of cloud and the lack of serious enterprise adoption of public cloud (versus data center and private cloud adoption) and I’ve reached the conclusion that the public cloud is: 1) ideal for new (or smaller) apps that have been virtualized and were intentionally created for the public cloud; 2) ideal for a fascinating debate about what constitutes various flavors of XaaS; and 3) is not ideal for existing, multi-tier apps that have not been modified or virtualized.
I also think that cloud migration might be the 900 pound gorilla that doesn’t get included when cloud costs are discussed, in addition to the usual secondary considerations of how security, data integration and governance requirements will be addressed. Most of the cloud costs covered by the press (as the costs of cloud computing) are the ongoing costs for apps that have already been migrated. There is very little, if any, data on the cost of migrating traditional (and especially multi-tier) apps into public or private clouds.
Three Reasons Why the Private Cloud Won the Enterprise
1) The cost savings and efficiencies for public cloud are highest at low or occasional levels of consumption and for smaller apps. There are some high profile examples of larger apps (specifically designed for the public cloud – like Netflix) that have been successfully deployed into the public cloud for continuous use. There are similarly high profile examples of apps that have been migrated off of the public cloud (for large, predictable workloads) including apps specifically designed for the cloud. Some of them are deploying hybrid clouds, opting for agility, efficiency, protection WITH control.
2) The public cloud has not yet evolved to a state whereby enterprises feel that they have the same level of security, control and management compared to private clouds and colocation. In a previous blog at Archimedius (Putting the Public Cloud into Perspective) I compared public cloud revenues (<$4B) with 3rd party hosting and colocation (>$20B) and overall IT spending (>$3T) to help put the public cloud buzz in perspective. Security, control, compliance and governance are frequently at the top of the “why we aren’t deploying apps in the public cloud” survey list.
The third factor hasn’t been discussed much outside of the cloud migration industry. I think it is a profound and, ironically, underserviced point.
3) The costs and risks of cloud migration and integration (including authentication, Active Directory and LDAP) are very significant gating factors. Yet there is very little third party data regarding cloud migration processes, risks and costs on any basis. Before our recent Hybrid Clouds: so Challenging yet so Promising webinar last week we approached a wide assortment of cloud migration and IT pros (as well as respected industry analysts) about the costs and time involved with migrating multi-tier apps into the cloud.
We heard about a wide range of “typical” projects, starting at about $100k for smaller, simpler apps with minimal database services to more than $10M for a larger, complex app with robust database services. A few firms independently suggested an average project size of about $300k for an app that requires about 60 servers. One firm had an average deal size closer to $100k but based on much fewer servers and less service complexity. Keep in mind that many of these projects were simply private cloud projects because the public cloud seemed even less feasible.
The irony was overwhelming: in between the increasingly complex and costly data center and the agile, efficient and available cloud is a mysterious ocean of manual processes, tools and billable hours that becomes ever more expensive for ever larger apps. The cloud perhaps requires even more risks and expenses simply to arrive and depart, which translates into an even weaker control story beyond commodity IaaS. Remember that 70s comedy show Gilligan’s Island?
It is time for the cloud migration service industry to embrace automation, not simply rely on outdated, reinvented tools, scripts, virtualization vendor lock-in and billable hours to enable hybrid cloud. As I type I can hear “The Gilligan’s Island Theme Song.” Here is a sample in case it may trigger the jingle:
So this is the tale of the castaways,
They’re here for a long, long time,
They’ll have to make the best of things,
It’s an uphill climb.