Buying and Integrating the Right Hardware for Cost- and Performance-Optimized Private Clouds

Hardware changes fast – almost as fast as software does. Best practices and rules-of-thumb change radically over just a small handful of years. So picking and integrating hardware for a new cloud build is challenging – even for engineers with relatively recent experience.

The risks are real: sub-optimal hardware strategy can easily end up compromising the cost, performance, and long-term flexibility of the cloud you’re building. And temptations to wrong-think abound: not least being the desire (common in engineers) to get as close to ‘state of the art’ as possible – especially with stuff like hardware, that’s purchased infrequently.

This article details some of the most important changes in datacenter server hardware over the past few years, and helps update the playbook for hardware selection and planning. Read it to learn:

Why now is the right time to switch from dual-CPU to single-CPU server designs.

How to right-size memory for optimal performance.

Why SSDs are now the right choice almost everywhere (but the cutoff point beyond which spinning disks still play a cost-effective role).

Sizing servers for optimal performance and platform resilience.