Give Me a Layer to Stand On, And I’ll Move You to the Cloud
Posted by Bob Warfield on April 14, 2011
This post is on behalf of the Enterprise CIO Forum and HP. This is my first in a series of posts sponsored by Enterprise CIO Forum and HP, thanks for the sponsorship!
Apologies to Archimedes, who is supposed to have said:
“Give me a place to stand and I’ll move the Earth.”
He was speaking of leverage and the idea that a great weight can be moved with very little force given the right amount of leverage. Leverage is what IT needs to make an orderly progression to the Cloud without having to expend too much force.
Where software is concerned, leverage often comes in the form of finding the right infrastructure layer from which to effect the desired transformation. A good layer may come in the form of some standard that becomes the Rosetta stone whereby lots of different implementations are made equal and customers have more choices. It may come from a particular abstraction that is powerful enough to sit on top of formerly quite different paradigms and make them all work alike. Leverage may come in the form of a new box of some kind that insulates one set of connections from the other, once again making the implementations on the other side of the box equal.
In order to move to the Cloud in an orderly and proficient manner, IT still needs the right layers to stand on. Cloud providers like Amazon have done a little bit of this work, but there is a lot more still to be done. You can see the difference in a service like Amazon’s, that looks very familiar to someone with at least a background in virtualization, versus a service like Google App Engine that forces you to more radically change how you think about infrastructure. A good Cloud layer will minimize that disruptive thinking, perhaps at some cost of performance, in order to achieve greater agility.
Ideally, the right Layers meet IT halfway to the Cloud, forcing them to change much less than a direct jump into the Cloud, and facilitating the process of making different infrastructure variations look the same. HP’s Hybrid Delivery is all about creating the right methodology and services to enable IT to think about their own data centers, private clouds, and public clouds as similarly as possible, as John Dodge points out over on the Enterprise CIO Forum. Dodge views the right layers as facilitating agility, and that’s exactly what a good layer ought to do.
The next step beyond methodology and services will be technologies that act as layers. IT will be able to refactor their infrastructure into components that are best left in the corporate data center, components that need a private cloud, and components that can thrive in public clouds. The test of the best technologies will be how agile and transparent this refactoring can be, as well as the completeness of the new layer in terms of solving the key problems:
– Security and Authentication across the different infrastructures.
– Latency issues that will arise when some data and API’s are leaving the data center and picking up a much more expensive round trip cost due to the latency of the Cloud.
– Management and Monitoring: How do we make it easy to do these jobs in the same way no matter which infrastructure is involved?
There are many more dimensions such Cloud Refactoring Layers will need to address, but those serve as a reasonable framework to start a discussion. As mentioned before, an appropriate Layer could take many different formats. Imagine, for example, a Cloud Appliance. Instead of installing a rack of blade servers, suppose you could insert a device in your rack that talked to servers in the Cloud and made managing and using them look very similar to having the physical blade servers right in your data center. What an interesting device this would be. Imagine a virtual Cloud “server” that acts as if it has 128 cores (or however many its appropriate to share per network connection in your rack based on the latency and bandwidth capacity of your Cloud access), terrabytes of data, compatibility with the management tools you know and love, a secure bulletproof connection to its Cloud backend that couldn’t “leak” into your network, and so on. Yet the device would sit there in your rack taking very little space and power. Consider it a Cloud “Force Multiplier” for your data center.
You wouldn’t have to entrust anything too sensitive to your Cloud Appliance. Perhaps it simply allows you to offload some capacity from your Data Center to make room for more mission critical apps. You could envision application dedicated versions of such an appliance aimed at apps like Mail Servers, Web Servers, or Sharepoint and other Social Apps. Or, perhaps the appliance would be tasked with doing backups for PC’s and the non-critical servers. Why mess with tape and other physical media when you can get multiple physical location redundant backups very easily from the Cloud? Such an appliance would be exactly the kind of leveraged Layer I’m talking about. It would make it easy for IT to start shifting apps to the Cloud without undue strain.
Such appliances already exist and are referred to as “Virtual Cluster Appliances.” Expect a lot more to develop along these lines.
This post is on behalf of the Enterprise CIO Forum and HP.