Rightsizing Your Cloud
There’s a perception that public cloud vendors have bottomless reservoirs of capacity. It doesn’t matter how much a business grows, how many additional users sign on, or how many petabytes of new data gets generated and stored – there’s always going to be plenty of additional space and capacity to go around, which seems to auto-magically appear out of nowhere when called upon.
Of course, there’s a lot more than magic involved – to meet constant and unpredictable demand, public cloud vendors maintain gigantic, industrial-sized data sites covering acres of land in many locations around the globe, with more coming online every month.
With private clouds maintained within corporations, it’s a different story. Cloud enables enterprises to better distribute their existing capacity, on an on-demand basis to those who need it. However, private cloud resources tend to be finite – at least until the next budget cycle, when more capacity can be purchased and installed.
Hence, for organizations with enterprise private clouds, there is a need to engage in the delicate art and science (it is both) of “rightsizing” resources to keep their ever-growing, ever-changing user bases online and productive. Rightsizing encompasses the following abilities:
- To automatically provision on-site IT resources as workloads demand
- To automatically “cloudburst” to other clouds if workloads exceed current capacity
- To grow with the organization, and predict, plan for and be ready for future workloads
- To monitor, respond to, and rapidly troubleshoot performance issues – preferably before they even happen
- To support applications and data on the most cost-effective platforms
- To enable self-service access by business users for applications and information whenever it is needed
The “art” of rightsizing involves working closely with business end-users. Unlike their public cloud counterparts, private cloud providers may have more limits to their scalability – in terms of I/O throughput speed, processing power, and disk storage space. Thus, compute requirements need to be communicated to the cloud team.
At the same time, private cloud managers have an important advantage that public cloud providers lack: an intimate understanding of the business, its IT requirements, and with it, a keener sense of future requirements. Internal end-users groups tend to be a known quantity, and their members may even be on a first-name basis with the technology platform providers.
The “science” of rightsizing involves calculating peaks and valleys in demand, and having mechanisms in place to handle both extremes in a cost-effective way. There is no guaranteed surefire way to accurately predict how much capacity application workloads may chew up as time goes on. Thus, private cloud providers may also need to have a “Plan B” established that ensures uninterrupted, high-performance service in the event of workload spikes. A hybrid approach may be called for, incorporating other cloud resources – such as external vendors’ cloud services – to assure continued performance.
With this new model of computing on the table, there will be a positive impact on IT budgets as well. With both private and public cloud resources at their disposal, IT planners and leaders can moderate IT spending, providing for enough capacity to run typical workloads, and only paying for additional IT resources for the temporary times they are needed.
For more thought provoking cloud management insights visit vmware-erdos.com.