IT has been a critical part of business response to the pandemic, Oostveen observed.

The first phase was all about ensuring business continuity, often by enabling people to work from home.

The focus was on productivity, for example by providing virtual desktop (VDI) access.

This led to such a large and sudden upswing in public cloud use that not all providers were able to live up to their promised level of elasticity, he said.

That phase is largely complete.

In the second phase, attention has turned to efficiency and to the use of analytics and AI.

However, organisations are suffering from a degree of "bill shock" as the invoices for their greatly increased public cloud use flow through, and this is leading to conversations about repatriating data to privately owned (or at least privately controlled) infrastructure.

In the shorter term, it's important to shut down instances that aren't being actively used, in order to control those bills.

On the other hand, the shortage of cloud resources mentioned above has led some organisations to keep instances running to ensure they will be available whenever they are required. This 'hamstering'...

 ...leads to even greater resource shortages. It also means higher bills, but that's a trade-off against the opportunity cost of not having the resources when you need them.

The realisation that public cloud isn't infinitely elastic is leading to some changes of plan, he said.

So Oostveen offers the following points for consideration:

• Applications built on a PaaS aren't portable. So if portability is important to you, make the appropriate architectural choices.

• You need a hybrid cloud. Furthermore, you need a hybrid cloud that extends to the edge. So if you don't have one, get one.

• "IT is sick of silos" but as fast as we pull them down, we build new ones.

Organisations are best served by choosing technologies that are as open as possible, but that means making the right design decisions, he concluded.