One conversation I’m still commonly having (weekly if not daily) is something similar to this:
What are the perfect vCPU to pCPU ratios that I should plan for and operate to, for maximum performance? How do I account for hyperthreads in that ratio?
The answer?
There is no common ratio, and, in fact, this line of thinking will cause operational problems. Let me tell you why.
In the past, we’ve used guidelines like 4 vCPUs to 1 pCPU (4:1) or even as high as 10 vCPUs to 1 pCPU (10:1), but this approach was based on an often unspoken assumption—those workloads were basically idle. Many organizations started their virtualization journeys by consolidating those systems that were easy to do so, and it was not uncommon to see very high vCPU-to-pCPU consolidation ratios.
Thus, consolidation ratios were born and became a foundational capacity planning construct for virtual environments. There was fierce competition over who could achieve a better consolidation ratio. Technologies like Intel® Hyper-Threading and AMD simultaneous multithreading (SMT) were introduced to provide better consolidation value. Now the ratio conversation became more complicated as you adapted it to logical thread counts instead of just physical cores. Large spreadsheets became the new operational dashboards used to manage capacity.
This approach to capacity planning, operations, and reporting, while once useful, is now outdated. The rapidly increasing churn rate of customer environments, coupled with the growing size and resource demands of virtual machines, has made it unsustainable. Moreover, with the widespread adoption of “virtual-first” policies, many customers no longer have the opportunity to profile application performance on physical environments before virtualizing them.
So if we can’t predict what will be virtualized, what its requirements are, or how long its lifecycle will be, we can’t create a static ratio for the commitment of any compute, memory, network, or storage.
Instead, we need to “drive by contention.”
By this, I mean we need to invest in pools of resources for application owners, so our new model becomes closely monitoring those pools for contention—which indicates the pool cannot support any more applications—and then growing them as required. This presents a new set of challenges that teams must overcome and master, especially with modern processors which can scale up to very high core counts.
We have designed VCF and its management tools for this purpose, not just for compute. At the platform layer, vSphere supports large clusters of resources that are dynamically balanced by services like DRS to mitigate the effects of contention over an application’s lifecycle. The VCF Operations suite monitors applications and pools of resources, letting you know when there’s a performance issue or when you need to manage capacity. This type of model allows the effective consumption of resources, giving you the best consolidation ratio, while ensuring application KPIs are always met. These results can’t be obtained with a static ratio.
Therefore, to escape the limitations of static ratios, ensure efficient hardware consumption, and adapt to the ever-changing business landscape, our operational models and toolsets require an upgrade. This upgrade must incorporate an understanding of advanced concepts, such as logical CPUs not always being equivalent to physical CPUs (as in hyperthreading) and the necessity of right-sizing virtual machines. The crucial factor for reducing risk then becomes how quickly you can respond to performance or capacity issues.
When quick response times aren’t feasible, adopting a conservative 1:1 vCPU to pCPU ratio (without considering hyperthreading’s logical CPUs) offers a safe initial approach. Over time, as your organization matures and adopts more sophisticated tools and processes, this ratio will organically improve. Its final optimal value, however, will be distinct for every organization, shaped by their specific mix of applications, technology stack, and operational maturity.
To summarize …
A 1:1 ratio gives you maximum performance at the highest cost. But in the world of doing more with less, learning how to “drive by contention” will ensure you maximize both performance and investment. VCF is designed to meet these challenges.
