The following is a contributed article by Ayman Gabarin, Senior VP EMEA for software-defined datacentre infrastructure firm Cirba.
A much-abused term when it comes to the world of IT “synergy” neatly encapsulates how, by combining technologies, we can build solutions able to deliver a lot more than the simple sum of their parts. As with handheld devices and wireless networking, for example, or when data encryption is paired with biometric authentication. And now we can add one more to the synergy hit list - the use of advanced analytics to extract greater value from all manner of complex systems including hybrid cloud infrastructures.
Fast becoming the flavour of the decade, analytics are being employed almost everywhere, primarily to make sense of the huge volumes of data being generated by modern IT systems. But that’s not all analytics can do here. The same technology can also be used to optimise the underlying platforms that are the backbone of IT with the aim of making them more agile in the face of rapidly changing business demands, regulatory requirements and so on. Hyperconverged platforms, traditional servers, storage and network systems can all benefit, with the hybrid cloud the latest to reap the rewards of this synergistic pairing.
The hybrid conundrum
You could be forgiven for thinking that there isn't much that analytics can do to add to the hybrid cloud equation, especially given the many column inches dedicated to the benefits of this approach to IT provisioning. That is, benefits such as being able to take advantage of scalable on-demand public cloud services to host business critical applications yet still mix them alongside others hosted on more secure private cloud platforms.
However, you would be wrong as, although fine in theory, the hybrid cloud model can throw up plenty of practical problems, not least when it comes to working out what workloads to put where.
The problem here is that enterprises moving to a hybrid infrastructure are, of necessity, finding themselves forced to change their approach to capacity planning. They’re moving away from a simple pre-purchasing model based, typically, on an extrapolation of current demand, to one that is dynamic and able to react quickly to peaks and troughs in that demand by taking full advantage of always available public cloud capacity.
This has not been an easy adjustment with many companies erring on the side of caution and continuing to skew investment towards their private infrastructure, even if it means missing out on what AWS, Azure and other public cloud services have to offer. Others, throw caution to the wind and opt to migrate as much as possible to the public cloud regardless of whether they need the capacity or agility it provides. Either way, these business-critical decisions are based largely on guesswork with the end result being costly over-provisioning and inflexible arrangements which simply fail to deliver the benefits the hybrid cloud approach clearly has to offer.
A more measured approach
The good news is that by factoring analytics into hybrid cloud provisioning, you immediately make a change for the better. Simply put, analytics takes the guesswork out of the equation, replacing it with certainty, based on an understanding of both the technical and business requirements of individual application workloads. This is an insight that makes it easier to determine which workload to host internally and which to move to the public cloud, which workloads to mix together, which to keep apart and so on.
For example, if you had an application with a peak workload every morning and more limited demand throughout the rest of the day, hosting in a large public cloud container would be inefficient and costly, resulting in expensive resources being left idle most of the time. Without analytics this would probably go unnoticed. With analytics it would be flagged up and you could opt to host the application in-house or alongside others able to make better use of those resources later in the day.
More than that, by monitoring workload patterns and applying predictive analysis models it becomes possible to not only react to changes in demand, but anticipate and optimise workload placement before those changes occur.
Knowledge is power
One small hurdle to overcome is the need to cope with the wide variety of platforms and technologies which, by its very nature, go to make up the typical hybrid infrastructure. Still this can be done, with SaaS the best approach here, enabling analytics to be added in a non-invasive, platform independent and scalable fashion. Plus if you’re considering the migration of applications to the public cloud, why not add another in-house to help optimise the process?
Another requirement is the ability to act on the insights made available by applying analytics to hybrid cloud deployments. The ultimate goal is the ability to automate workload placement based on specific recommendations provided by analytic models. This, in turn, requires organisations to have some form of automated provisioning, which many do for internal placements although, for many, is still manual when sending workloads out to public cloud platforms at this time.
What is critical to all applications of data analytics, not just hybrid cloud management, is that they can be leveraged by unskilled users who are also given very specific direction as to what resulting actions should be taken. That said, there is still a need to build the expert knowledge of data scientists and business analysts into the modelling tools themselves to fine tune them to the unique policies and considerations of an individual organisation. This is where the real power of advanced analytics can come into play, especially organisations with special needs such as those in highly regulated industries.
The bottom line
According to Gartner more than $1 trillion in IT spending will be directly or indirectly dependent on the shift to cloud computing over the next five years. That’s a lot of money and an investment that needs to be based on a true understanding of the needs of the applications involved rather than guesswork. This is an understanding that can only come from applying best-of-breed analytics to determine where to place and optimise workloads within a hybrid cloud infrastructure, to deliver much needed synergy and better meet utilisation requirements, service levels and operational policy requirements while minimising infrastructure needs and costs.
PREVIOUS ARTICLE«Can cloud rivals catch up with AWS?