Enterprises that leverage cloud computing see a few key improvements, including operational cost efficiency, speed to bring data and applications into production, and the value of business ability.  However, we’re still very inexperienced with the use of cloud-based systems.  While there is a great deal of productive use of this technology, there are instances where we’ve fit square pegs into round holes.

These mistakes are not easy to spot.  Many in enterprise IT consider anything that works in the cloud as a clear success.  However, in many of these cases, these solutions are much less than efficient and effective, and could end up costing more money in the long term than expected.  Thus, the value of cloud is significantly diminished.

So, what does enterprise IT need to work on when it comes to cloud computing?  Here are three that top my list.  I’m sure you have your own.

1.    The tendency to leverage private clouds, no matter the requirements.  

The use of private clouds is indicated in about 50 percent of the problem domains that I see.  In those instances, it just makes sense to keep the applications and the data local, while also leveraging the functional advantages of cloud computing, including self- and auto-provisioning, elasticity, etc..

The problem is that many enterprises find the concept of private cloud to be an easy way to check the cloud computing box for their bosses, yet keep all data and applications in the data center.  They typically cite security and compliance issues.  However, overuse of the private cloud model means that much of the value cloud computing can bring remains unrealized.

Private clouds, which typically provide more efficient use of resources, still mean that you’re buying, hosting, and maintaining your own hardware and software.  That means that you do not reduce the overall costs of computing.  Indeed, in many cases that I see, the use of cloud computing actually makes things more expensive.  More redundant resources are added to the data center that have to be powered and managed.

2.    The tendency to leverage a “single brand” strategy.

There are IBM shops, Microsoft shops, and Oracle shops, so why not AWS or Rackspace shops?  Relying on a single brand of technology, including cloud-based platforms, typically means you’re not leveraging best-of-breed solutions.

Those who are moving to public clouds often focus on a single public cloud provider.  They forsake all others because they want a single-brand solution to both simplify their cloud computing solutions, as well as provide “a single throat to choke.”

The notion that sticking with a single cloud brand is a good idea is just not based upon sound logic.  Some public clouds are good at storage, others good at compute, still others are good at providing database-as-a-service, and a few specialize in big data.

Chances are, if you do your homework and testing, you’ll end up with a multi-cloud solution which will provide you with the most optimized solution for the money.  It might be more complex, but also worth several million dollars a year in savings.  The throat to choke could be yours, if the CIO figures that out.

3.    No consideration for performance, governance, and security. 

While you would think that most who deploy public clouds think nothing but security, many implement 1980s style security that just won’t work in the cloud.  The concept that security means locking everything up to everyone and everything just won’t scale.

You need to consider the fact that cloud computing is a complex distributed system.  Complex distributed systems should have fine-grained approaches to security that allow for the configuration and re-configuration access rights to resources.  This typically means identity management types of security, which have been catching on lately with cloud computing security solutions.  Most enterprises don’t think about identity management issues, and thus end up with a security solution that either limits the capabilities of their cloud solution, or makes the cloud solution vulnerable to breaches.

Governance is a different beast that also gets neglected.  You need to manage resources using well designed governance technology, such as cloud management platforms (CMPs), or service governance systems.  These are designed to use policy-driven approaches to protect the use of resources and APIs, and control them using a single set of interfaces.  Without this technology, complex cloud deployment can’t be properly managed, and may not scale.

What’s interesting about the issues outlined above is that most enterprises don’t consider them issues at all.  They will drive forward with cloud, a bit clueless about approaches and technologies that can make their cloud deployments a magnitude more cost efficient, and thus additive to the value of the business.  The ‘cross your fingers and hope’ approach to technology implementation usually doesn’t meet with much success.