Cloud Bursting Not Ready for Prime Time

Arthur Cole
Slide Show

Choosing Your First Cloud Application Initiative

Questions you should ask to help determine which cloud application path you should pursue.

It all sounds great on paper: an integrated physical/virtual/cloud environment in which resources are spun up at the drop of a hat allowing enterprises to consume exactly what they need and not a bit more. In such a world, "cloud bursting" is as common as apple pie and CIOs never have to worry about over-provisioning or under-performing.


So far, though, the reality is a lot different. Enterprises are indeed becoming more agile and data environments more dynamic, but the idea of automatically bursting excess loads onto the cloud is still largely a myth.


For one thing, notes ZDNet's Manek Dubash, moving data from one set of resources to another still takes a fair amount of time. So if you want the flexibility to suddenly shift data spikes onto the cloud, you'll need to have things like applications, storage and bandwidth standing at the ready, which costs money. It's probably not as expensive as maintaining your own backup infrastructure, but it still represents a set of capabilities that you're paying for but not necessarily using.


Even with cloud resources standing by, most enterprises will encounter any number of hiccups when trying for the kind of instant hand-off that cloud bursting promises. For that, you'll need a new kind of automation designed specifically for local/cloud harmony. Bright Computing's Bright Cluster Manager 6.0, for example, features a data-aware scheduling function that allows enterprises to extend their on-site server and cluster architectures onto the Amazon EC2 cloud, essentially allowing the entire environment to be managed as a local resource. With automated job initiation and data transfer, the platform also eliminates the need to proactively monitor and manage data at every step.


However, these kinds of tools are only useful in the cloud if the local data environment is in reasonably good shape. Far too often, according to CBR's Jason Stamper, cloud bursting falls victim to poor management and control in the data center. If IT has a poor grasp of its own resource management, capacity planning and automation processes, it will be nearly impossible to gauge the efficacy of any cloud solution.


One of the lynchpins to cloud bursting, of course, is the hybrid cloud, which mixes both public and private resources into a cohesive whole. The problem here, according to InfoWorld's Doug Dineley, is that the hybrid cloud is more of a marketing creation than an actual solution at this point. For one thing, no one is providing multi-site management in their cloud platforms yet, so even if internal and external clouds share common toolsets and other features, they still must be managed separately. And despite advances in network virtualization, the WAN is still too slow for truly seamless migration.



So is cloud bursting a pipe-dream? Just another unrealized expectation to fall by the wayside as the new technology's limitations begin to emerge? Hardly. Most of the issues blocking the path stand a good chance of being resolved as new generations of networking and management platforms come out. It's doubtful that we'll ever see instantaneous, fully automated bursting, but at least we'll have the means to prep the cloud should loads start to approach predefined thresholds or if peak operations fall into predictable patterns.



Add Comment      Leave a comment on this blog post

Post a comment

 

 

 

 


(Maximum characters: 1200). You have 1200 characters left.

 

null
null

 

Subscribe to our Newsletters

Sign up now and get the best business technology insights direct to your inbox.