Consensus On Hybrid Cloud

By now, you may be tired of all the industry hype touting cloud infrastructure. Amazon, Microsoft, Google, Rackspace, Gogrid, Flexiscale and many other companies are actively marketing the cloud nirvana available to all for just one monthly payment per user after another. This software, platform or infrastructure is ready to deploy at a moment’s notice, scaling hurdles are a thing of the past, and maintenance is someone else’s concern. All these benefits are easy to sell to small businesses that need software to just work because they do not have staff to perform installs, forklift upgrades, deployments or break-fix.

The last concern of any of these vendors is that cloud computing is an ambiguous term. No definitive definition exists. Everyone knows cloud computing involves the internet, but beyond that, things get cloudy. Is the vendor talking about distributed computing or not? Is the vendor talking about virtualized platforms or not? Is the vendor just using the term to mean hosting? Is proprietary software involved? Since the answers to these questions vary on a per vendor basis, cloud computing is a nebulous concept.

However, this ambiguity actually plays into the hands of the vendors. If they can claim they have a secret sauce recipe and their software just works, then no one can criticize their design. Take a common customer relationship management (CRM) tool like SalesForce (SFDC). When a customer uses this service, it is assumed that SFDC has redundant servers that are ready to take over should one of their production machines fail. It is also assumed that their company conducts regular backups in case there is software corruption, service interruption, or hardware failure. It is assumed that their software is secure and their development teams are proactive about responding to bug reports. But these assumptions are only that because SFDC does not share their infrastructure architecture or provide a tangible check on how some of these tasks are accomplished.

As a small company with a small customer base and a small amount of internal intellectual property (IP), taking a risk on a cloud provider may make economic sense. When a company has been in business for several years and has amassed a significant amount of IP, this risk should be re-evaluated.

A customer is meant to take any cloud provider at their marketing word that given the vendor’s core business, they are better at the service they advertise than the customer could be. As long as the customer cedes control over their intellectual property, the cloud vendors can solve many technological and infrastructural challenges. The danger, however, is that ultimately the vendor is not responsible for your data. Which means, regardless of all of the assurances a cloud vendor has in their marketing copy, if your data is lost, leaked, corrupted, or stolen, the vendor is not liable.

Over the next decade, as WAN speeds catch up to current LAN speeds, the list of services that is not offered by cloud providers will shrink. There will continue to be exceptions on that list because of network limitations: latency or network bandwidth may prevent real-time applications or storage-intensive applications from moving into the cloud.

This does not mean that the overwhelming majority of companies will outsource their IT into the cloud. In fact, except for small businesses, companies will adopt a hybrid approach by hosting their own private cloud using the latest advances in on-demand deployment within their virtualized environment and relying on public cloud companies for non-critical services or spill-over utilization. For the majority of companies, control over their IP is worth the effort of keeping in-house infrastructure. At the same time, using cloud providers for certain less critical services makes perfect economic sense.