Altan Khendup

It is always asked in cloud computing circles. Heck even outside of these same circles. "What is cloud computing?". There are good definitions such as those on wikipedia and from standards bodies around the world. Yet many of these speak to the technical professionals and rely on these individuals to explain the concept to their customers resulting in very ambiguous and sometimes incorrect expectations.

One of the best explanations given for cloud computing happened to be at a recent local meet-up where it was described as "making infrastructure available to the masses". Regardless of whether or not cloud computing in it's current form is at the application level, infrastructure or platform, they are in essence doing just this.

As a result some of the promises of cloud computing are very apparent:

• Data center/infrastructure resources can be dynamic; being allocated and de-allocated based on need


• Speed in deploying what is needed, where it is needed


• Simplified representations of complex resources and operations to make better decisions


• On-demand resource ordering and provisioning for various customers

And the list goes on. Yet when looking at any typical business scenario or even personal ones, the questions about how to best use it are very apparent:

• Do I put everything on someone else's solution? Public/vendor clouds.


• Do I create my own solution? Private clouds.


• Do I mix up the two? Hybrid clouds.


• How can I use a portion of A, mix it with B and C without jumping through a lot of hoops? Cloud interoperability

Some of the challenges are also obvious such as security, integrity, availability and the like. After all everyone has experienced an outage like email or phone service and knows how inconvenient that can be, and know how potentially fatal it can be for a business.

Yet making the data center infrastructure more readily available has some challenges that are ages old. A classic one is monitoring. Everyone has seen the little monitoring widget that comes with their personal computers - CPU, memory, disk, network, processes, etc. Yet how many people actually use them aside from the technical personnel? Very few. Even for technical personnel interpreting this information in aggregate and mapping it against the business transactions is highly fluid. Let's say one is looking at 12 computers with 4 CPUs each. That is at least 48 CPU monitors, 12 memory monitors, 12 process monitors, 12 disk monitors, and 12 network monitors. This is pretty typical of a small computing environment. Now let's scale that up to a data center which has let's say 500 computers with the same 4 CPUs. 2,000 CPU monitors, 500 memory monitors, 500 disk monitors, 500 network monitors, 500 process monitors - that is a lot of monitors. That is for 1 data center, now do that for say 12 data centers. And you can see the issue.

Even armed with today's most complex and best tools from vendors and open source projects, interpreting this much data for various business applications has been very difficult for most organizations who have trained personnel. Imagine the cloud vendors or even your own cloud in trying to map tackle this challenge. Throw in the required dimension of determining meaning for varied usage patterns based on varying applications and this can be very difficult. Or even simple troubleshooting of a slowdown or similar issue for a single business activity. Many of the tools are not designed to deal with the size, speed or flexibility of clouds. They are too slow, too ponderous, or requirement too much manual intervention to be useful.

This is but one sort of challenge that is being addressed by the cloud community. As cloud move inexorably into being more mature, there is a lot of innovative thinking into tackling these and other challenges for consumers of clouds.

As the cloud computing landscape changes, it is important to come to a good understanding of what clouds can offers in their current state and adjust it as time goes on. And I like simple, easy-to-grasp definitions. It makes explaining things to customers a lot easier.