Running data centres efficiently is always has been a big challenge for IT managers. But there is one aspect which provides some relief to them – expansion in choice of how to run them. Today, they have options such as using server virtualization and cloud computing.
Today, server virtualization dominates as a strategy to reduce data centre costs, but cloud computing is also catching up fast. Cloud computing in a nutshell is delivery of services such as computing, storage etc. as a service.
But the basic question arises, why to go for cloud when one is already getting cost benefits while using techniques such as server virtualization. The key difference, experts point out is the automation and orchestration provided by cloud computing which is missing in server virtualization. The layer of abstraction imposed between applications and servers by cloud computing automates many tasks.
So, there is always confusion in terms of how to decide where to run applications and which ones. Here are a few simple pointers which can help decide this.
Data centers, Cloud or In-house : Key points to keep in mind
Some key points which should be considered while evaluating whether to use cloud computing or not:
Which applications and why ?
Clarity is required in terms of which applications can run on cloud and why? Do these applications need cloud because the internal data centres cannot run it efficiently, or is it because of storage requirement or others. For example, applications whose computing requirement change significantly over time (have lots of peaks and troughs) are good candidate for cloud because the infrastructure required to run their peak load remains idle in other situations. It is important to arrive on the list, with as much clarity as possible with solid reasons.
What will cloud option entail ?
What are the expectation from the cloud in terms of storage, computing power, uptime etc. ? Is there any vendor meeting them within the budget? Cost always plays a crucial role in short listing options. Some options will just be too costly to implement at that stage.
Security : What level of security is required by applications to run? Some applications carry very sensitive data and thus how is data handled while using cloud vendors is very important. In most cases, cloud computing is not preferred for sensitive data. If cloud computing has to be chosen it should be internal clouds. Public clouds should be avoided altogether for such applications.
How easy will it be to go cloud ?
It is not always easy to go cloud and it may require significant IT effort. Applications need to be virtualized before they can be put on cloud and if they are already not, effort increases.
Attention to service level agreements
Due attention should be paid on service level agreements. If the services expected to run applications per requirement are not provided, it is best to search for another vendor keep them in-house.
There should be a cost benefit analysis considering short term and long term benefits. Firms which just see short term gains of losses often make mistakes as scenario may change in long term. It is important to consider all kinds of costs while doing this analysis, e.g. infrastructure costs, human resource costs, management time etc. Also consider the utilization rate of the in-house infrastructure (which generally keeps fluctuating).
In the end, one can say that irrespective of the benefits offered by the clouds, the firms which have invested heavily on building internal data centre infrastructure will not just shut them down and shift to cloud. The shift is expected to be gradual. In many case firms will also build internal clouds to handle sensitive data and applications. But one thing almost everyone agrees upon, preference for cloud will increase simple because it is more efficient and cost effective way.