Getting to grips with workloads in the cloud

Without workloads the cloud is just idle. This perhaps seems very obvious to all of us because wherever data is being processed workload is created by the servers used to manage and handle it. David Chalmers, Technology Director for Hewlett Packard’s Enterprise Business Group agrees: "“I guess this is true because all good technology ideas need workloads, but the market is looking beyond looking at ideas to examine workloads that offer business benefits.”

He argues that "without meaningful business applications the cloud is just an interesting idea”. So it only becomes beneficial when people move beyond real business usage. No matter what you think about the validity of the workload argument, it’s true that every organisation needs to ensure that their virtual machines are being fully utilised and able to fully process their workloads.

Program in resilience

CA Technologies’ cloud advisor, Gregor Petri, also believes that when you are moving data from one cloud provider to another you have to program in some resilience into the process. "After all you cannot go into the data centre and fix things if something goes wrong, and so all you can do is move your workloads to another cloud provider or to another data centre that is situated in another region", he explains. He adds that consumer cloud service providers like Netfix ensure that it is built into their applications when they are programming and developing them.

The trouble is that most organisations don’t program their applications from scratch, instead they are using ones that were either built before the existence of the cloud, or their applications came as standard packages. Petri says this were workload automation can offer the ability to move their workloads around as and when necessary. They might, for example, want to move them from being located internally to Provider A, B, C or from Data Centre A, B or C of a certain cloud provider.

Two things you can do
Andrew Bates, chief technology officer for transformation services at CapGemini says that work can be relocated to the cloud in two ways, and in both cases there are big opportunities and challenges. This is because not all workloads are currently suitable for cloud deployment. Yet they can be moved in the following manner:

• Dedicated environments can be provisioned in the cloud (static provision). So far this has been used largely for test and development environments but will increasingly be use for production in the future.
• Production environments can be expanded into the cloud dynamically as workloads demand it, and then contracted back. 

He believes that this is where the cloud is maturing, but it also encompasses a number of challenges. This is because not all workloads are in tune with being processed within the cloud, so cloud bursting or business services that transcend private and public cloud create many issues.

So when an organisation wants to move a process into the cloud, the data it needs also has to be transferred with it. Bates claims that this approach works well with functions like financial, oil and gas geophysical modelling. This is because there is a small amount of data and a lot of computing power behind it. The tasks that require the handling of a huge amount of data, or workloads that require a high level of security compliance, with which come regulatory concerns and considerations, tend to make things more complicated though.

"That’s because you can’t move all of the data into the cloud, and you might for security reasons be unwilling to do it too”, he explains. This is why he thinks that the network is both the enabler of the cloud as well as its limiter.

However, it’s the very attributes and the benefits of the cloud that are attracting customers. This includes the ability to gain rapid access to standardised services and its ability to offer a flexible commercial model.