Hooray for the human touch
October 8, 2021
Broadband networks and all that comes with them – CDNs, nodes, edge servers etc – were expected to fall over in the pandemic. The forecast of friction points, and outright failure, as traffic caused by WFH and more time to use entertainment, was so solid that the big streamers voluntarily cut their data intensity, trading some loss of quality for lower volume (a cynic might also point out that they also, thereby, cut their power bill).
But they needn’t have bothered. With almost no exceptions, networks worldwide held up pretty well. Sure, there were few live events but, even so; chapeau to the nets. Their resilience was in no small part down to the advances in real time monitoring, diagnostics, and even healing, provided by the participants in our annual survey of the test and monitor scene.
It is striking that, for many, the only major amber light on the dash was in the human side of the story: how do you win new customers, how do you coach new deployments, how do you recruit and train new staff, without human interaction? Of course, players in this field are more used to virtual working than most, but it has still been a major challenge.
Darn it. When will they iron out this last wrinkle in the virtual world? Every time a bad thing happens, what is the faulty component? As I write this Facebook and all its spawn have just come back online after six hours of global outage. They haven’t named a reason for the ‘calamity’ as yet, but it seems pretty sure it was a human making some kind of change to the DNS. And because Facebook’s greed, I mean constant search for economic and technological efficiency, had led it to lump all its services – Facebook, Instagram, WhatsApp and all – on one private net to speed up availability and cross-sell ads to your cross-matched data, when the public internet couldn’t find the top line Facebook DNS it couldn’t find anything at all.
There is something very comforting that when calamity comes, most likely it still isn’t an unexplained catastrophe in the machine learned code, nor a sinister invader from cyberspace, but an over worked techy, or an over-confident intern, or someone with just oversize fingers, plain old getting it wrong.