Costing Elastic Scalability with Hybrid Clouds


VMware has now announced its partnership with AWS to host vCloud workloads on Amazon hardware. That means we are starting to see the dream of a hyper-converged hybrid cloud become reality. This convergence brings the need for a new TCO and ROI calculation that dramatically differs from our previous models—so is consolidation ratio back in fashion?

When planning to migrate from a private cloud to a hybrid or public cloud, the most difficult task is the process of measuring and quantifying the cost. For larger organizations, this process can take months; for smaller organizations, the tools to quantify can be costly. Adding further complexity is the need to elastically move workloads back and forth across the platforms. VMware’s recent partnership announcements with AWS for hosting workloads in the cloud on dedicated, hyper-converged hardware calls for a new strategy in costing models.

Traditionally, we needed to look at how many workloads can reside on a server in the private cloud. Consolidation ratios were king. The mantra was “more cores, more RAM, more VMs” per host. This kept license costs low and maximized the utilization of resources we had purchased for our data centers.

When cloud computing came along, we stopped looking at consolidation ratio, as we did not know what other workloads were running on a host with us. We began to think about rightsizing workloads to reduce monthly cloud-hosting costs. This exercise focused on rightsizing each workload to its ideal vCPU, vRAM, and storage configuration. Density was less of a factor but knowing your true utilization meant you did not over-buy. This was a strange paradigm shift from the days of overcommitment, memory swaps, and CPU Schedulers.

Now, we are entering the third change in cost analysis. With the combined services of VMware and AWS offering compute clusters in the cloud with dedicated hardware, we start to turn our attention back towards density. Dedicated resources let us control the density and contention in our public cloud. Dedicated hardware means understanding our aggregate CPU GHz, aggregate physical RAM, and the maximums of throughput for storage and network. These physical limits become our boundaries in planning and pricing for the new hybrid / public cloud model. VMware vCloud on AWS includes the full stack of VMware SDDC services on dedicated hardware, hosted and managed by Amazon Web Services. Compute handled by vSphere, storage handled by vSAN, and network managed by NSX all translates into a power coupling of platform and services. But at what cost point does this model become practical? The exercise of building out an ROI / TCO model means planning for resources where you don’t yet know what you need or where they will run. This necessitates a true dynamic calculator.

To be continued…

Rightsize Your Environment Today!