Migration Planning and Workload Placement
More and more often, people are either actively migrating workloads to the cloud or coming up with reams of reports and binders of completed consultations about how to do so. The wealth of information isn’t comical or out of place- it is a large, important decision to make this kind of change and commitment, and will affect all levels of the business. At times, the meetings and various bureaucratic bargaining surrounding the initial implementation bring to mind the crowded control rooms and urgent, tense atmosphere of the early Apollo missions- as well they should.
Lurking behind the welcoming user interfaces and sleek Web 2.0-styled backdrops that cloud providers are known for are the very real specters of initial / ongoing cost and budget. Converting and expanding fractions of pennies into yearly costs, poring through template offerings, regions, zones, and more… if only there were some system that already understood the demand placed on my environment, and would intelligently select a template based not on an allocation model, but with a demand-driven approach in mind!
Of course, Turbonomic is that system and understands the complexities of your environment due to its n-dimensional approach to finding your desired state. With all the information at our digital fingertips, we would be remiss if we did not also use this veritable treasure trove of data to assist you in making intelligent decisions when migrating workloads to a cloud environment.
The largest mistake that most humans make when moving workloads to the cloud is to look at the current allocation of each virtual machine and select templates based on those values- this is commonly known as the “allocation-based model”. This model suffers from a very common flaw – it does not take actual utilization into account at all! The virtual machine that has 32GB allocated but has never used more than 4GB in its lifespan will, using this method, require a much more expensive template than is needed in order to live in the cloud.
At the time of writing, the difference in a template to support 32GB of RAM and one to support 4GB is more than $3000 annually – for a single virtual machine. Imagine the cost in both time and sanity required to research average workload, available templates on AWS/Azure, and perform these kinds of calculations for every virtual machine in your environment- and if you’re in a position where you would like to move workload around in multiple public cloud spaces based on value, add in several more orders of magnitude.
With Turbonomic, your workloads will be migrated to the cloud based on actual usage data and performance characteristics will be used to place workloads in the correct zone, with the correct template – defining “correct” as “scaled to the ideal template for each machine according to the utilization”, rather than allocation. In addition, Turbonomic can inform you how much of your existing, on-prem infrastructure can be repurposed or retired- an immediate source of savings, as any further on-prem expansion can be handled with existing hardware.
In the next article, we’ll be looking at the following plan, and how it will assist in your decision to expand to the cloud (all the best two-parters end with a teaser image!):