The Golden Understated Age of Infrastructure

The Golden Understated Age of Infrastructure

This year I marked an anniversary of sorts in that I have attended a decade of MWC events. And in those ten years, I have looked at many many hardware chassis, from proprietary “hot boxes,” “pizza boxes,” cPCI, ATCA, MicroTCA and myriad others.

And this year was in many respects similar but also somewhat understated. While the shift to the cloud has certainly lowered the spotlight on hardware and compute platforms, the truth is we have never relied more upon these platforms for common compute and storage. Still, when you look at the physical appearance, outwardly not all that much has seemed to change. But appearances are deceiving given these platforms truly represent quantum leaps in terms of hardware and software performance from platforms sold only 24 months ago.

I suppose this is to be expected if you apply the long-standing principle of Moore’s Law. However, I was unsure if this principle and continued performance breakthroughs would be enough to meet even our short-term cloud compute requirements. The concern I have is that we are now facing near zero millisecond latency budgets to accommodate 5G services and the realization of the dream of the tactile Internet offset with truly massive linear scale requirements. And then there is the notion of distributed clouds that 5G cloud RAN and other architectures like Mobile Edge Compute (MEC) will drive. While I certainly agree that pushing compute resources to the edge is crucial to achieving ultra-low latency targets, it’s hard not to be concerned by these massive loadings we are pushing to the edge, while maintaining the need to support traditional telco five-nines reliability performance targets. (See Cloud-RAN: Cost Savings Strategy or Key Enabler for 5G & IoT Use Cases?)

Moreover, it’s not just 5G and edge computing that are imposing these extreme platform requirements — security must be factored in as well. And without question, while applying analytics and policy is the key to mitigating the risk in the cloud, again the impact on software resources to execute these policies and support seamless VNF failover is immense.

Therefore, when I entered MWC17 for this tenth time, I was expecting that hardware and software wise, I would reach the same conclusion I reached at previous MWCs, namely that the latest, most advanced compute platforms of that era had nearly enough capacity to meet short-term compute requirements.

But this year was different. My observation was that the industry has quietly reached a point where the intersection of hardware and software has resulted in the creation of open carrier-grade-based server platforms that clearly exceed not only short-term requirements, but are now following a strategically sound development track that will ensure these forecasted quantum performance jumps can be accommodated.

As a result, even though 3GSM World Congress 2007, with its large number of platforms with integrated software on display, may have more appropriately represented the golden age of infrastructure, MWC17 in my mind will be remembered as the event in which server platforms — both from a hardware and software integration perspective — have in an understated fashion attained a truly impressive performance curve that could never have been anticipated in the bulging halls of the venerable 2007 Barcelona Fira.

Article originally published on Light Reading.