Different workloads perform differently on different cloud service providers. Enough so that it is prudent in planning to consider the optimal configuration and the optimal CSP for your solution. Consider this old word problem from years ago. One person can carry two buckets of water. It takes 5 minutes to fill the buckets and 5 minutes for one person to carry the buckets, empty the bucket and return to the well to get more water. The unit of work is one person and 10 minutes to move two buckets of water. If you have two people operating and assume they talk more because there is two of them let’s say, two people can move four buckets of water in 10 minutes and 30 seconds.
When will the two person team be ahead of the one person team in terms of units of work? It occurs by the 2nd trip or 30 minutes (8 buckets versus 4). Now you can in the end speed up the process in the wrong way (such as having a person filling the buckets for the person when they return from each round trip). But like CSP’s we have many options. The first and most immediate way to speed things up is to have two people moving the water. The next quickest way is to have more than four buckets and having a third person fill the buckets while the other two are walking. Of course, we can also have many more buckets and many more people to move this along even faster.
All of these are considerations in building out your cloud portfolio. For solutions that require speed and throughput you may select a different CSP than for solutions that require ubiquitous access (network and device), but don’t require speed. One of your selection criteria should be the value of the above equation. How much water does your workload need to move?
The next consideration is the reality of security. All the CSP’s I’ve worked with actually leverage a different security model. They all report issues the same way, but they use a different model both for evaluation of threats and protection of that. The more CSP’s you have, the greater your risk of missing something from a security perspective. This may be the single “why a cloud broker?” argument. The broker would connect to your enterprise and from there would abstract the cloud services you were connecting to. They may provide identity management services, and they will provide a unified security POS (Point-of-Sale).
Eventually, as the tools play catchup, the broker may even provide the great white whale of portability. The broker is giving you the ability to move your solution effortlessly between Cloud Service Providers without disruption of service or productive time lost.
With a broker, you can change the equation above. First off the broker will give you the ability to connect directly to the well without having people moving the buckets of water. So now you can pump water from the broker at a much faster rate. Secondly the security of the water is increased as it is under your partner’s control (broker) or your control longer. Finally, the broker gives you the ability to move your water source (eventually when the tools catchup) without having to dig a new well.
Many analysts project that organizations will have more than one cloud and more than one type of cloud going forward. hybrid cloud represents a mix of one or more cloud types and one or more cloud providers.
When customers ask me, I always tell them one thing. Every CSP you connect requires three distinct connections
In the end a Hybrid Cloud solution that supports a private cloud solution, and two or more public cloud solutions would have three of each of the connections above. Or if you implement a cloud broker one connection for each – and you can continue to add CSP’s forever without making more connections.
It’s why in the end Cloud Brokers make sense. Please excuse me for a moment while I go back to bailing the water out of my basement.
By Scott Andersen
