Cloud Lock-in and Pandemic Resilience

April 17th, 2020

As the world deals the COVID pandemic, it looks like some companies might be regretting the move to put their entire IT infrastructure in proprietary cloud platforms. When the times are good, sure, a large cloud bill doesn’t seem so bad. But what happens when cash flow halts to a stop and your IT-landlord doesn’t budge?

I fully expect this event to cause a shift in the upcoming years, as the next generation of IT operations starts to consider alternatives. The “everything in the cloud” mantra was sold en-masse to middle managers throughout the 2010s, with flocks of companies locking themselves into cloud platforms that impose painful migration costs.

I refuse to build anything that cannot be easily ported between cloud platforms. It’s just a bad idea. For example, S3 is good, because there are plenty of S3 providers, even FOSS and self-hosted, albeit perhaps with less functionality. Just don’t use the extra functionality. AWS Lambda is awful, because you’ve got only one provider. It’s not fear that Amazon might disappear in the near future (a far fetched event), but that I involuntarily grant Amazon a monopoly on running my software (while locking my users too in the case of FOSS, which is even worse).

Speaking of FOSS, we should mention that in terms of resilience any company that widely adopted FOSS in their daily IT operations is probably patting itself in the back; as millions of remote workers leave their office machines at work and leave behind their precious proprietary licenses LOCKED to their office computers, in some other companies employees simply download a copy of their favorite FOSS stack and continue work more or less unfazed.

The times they are a-changin’.