'Amazon has destroyed the unicorn factory' ... How clouds are making sysadmins extinct
Who'll build your data center if they all work for AWS, Google?
The rise of the cloud is wiping out the next generation of valuable sysadmins as startups never learn about how to manage data-center gear properly, a Pasadena tech biz boss has said.
The problem, according to Steve Curry – the president of managed OpenStack provider Metacloud – is that modern upstarts are going directly to cloud services like Amazon and missing out on the crucial experience of managing (and breaking) their own data center kit as they grow.
Finding people with skills in "large-scale operations at a global scale" is "really rare," he told The Reg in San Francisco on Thursday, and worried it may become almost impossible to recruit such techies.
"It's extinct now, here's why it's extinct – Amazon has destroyed that unicorn factory," he said.
Unicorns, he explained, are "those smart people [that] everyone wants but no one has ever seen one." They typically have experience managing large amounts of infrastructure and have a background in data center administration and suffer no fools gladly – much like our own Trevor Pott.
"Amazon killed that unicorn factory because people used to get money from a VC, they would have to go out and acquire hardware, put it in a data center, build that company's presence up technically from the ground up and then scale it and they learned how to scale it over time," he explained.
"No one does that anymore: they get their money from a VC, they go straight to Amazon, they put all their money there, they invest heavily in engineering or building apps or what is core to their business, and at that point infrastructure is not core at all because Amazon does that all for them so that skill level is not being replenished."
This pattern has played out at numerous startups, including ephemeral messaging service Snapchat which started life on Google App Engine and then hired away senior Google cloud employee Peter Magnusson to help it figure out how to move bits of its workload off the cloud.
Similarly, backend-as-a-service company Firebase started out life on a major public cloud but had to shift to bare-metal servers after it ran into performance issues.
"Cloud servers have a virtual network stack inside the hypervisor," the company's CTO Andrew Lee told us at the time. "I was surprised by how early [the problem] came – we had to move off much sooner than we would have liked."
Though investors and pundits will encourage the valley's startups to go to the cloud, it seems likely that the tech community is creating a problem for itself.
After all, Metacloud has had to hire people from Google at great expense to get the skills it needs to manage its own infrastructure, and not everyone can afford the digits to tempt an employee away from those ad-gilded confines. If all companies use the cloud, then the pool of available talent to create new infrastructure-specialist companies will shrink – and that strikes us as a bad thing. ®
Sponsored: Benefits from the lessons learned in HPC