We will not speak about compute and object storage; it’s obvious that they take the first two positions. But let’s talk about the next three.
At Hystax, we’ve recently conducted a broad survey with more than 400 respondents and got interesting results. I expected to see firewalls, big data services and cloud native databases – services that can fully utilize elasticity of a cloud. But I was not 100% accurate. So the top three are:
Relational databases (RDS, Google Cloud SQL and Azure Cosmos DB) – no surprise here as everybody uses databases, and, if the company is ‘born in a cloud’, there is a perfect sense to use the cloud native service. But… There are some companies which explicitly say that they run databases in VMs only as they don’t want to get into a ‘vendor lock-in’ trap.
34% of respondents use Lambdas for various tasks. Some survey participants run it for compute, some run clean-up scripts by schedule. It looks like the technology to execute some piece of code is highly adopted for different granular tasks.
Certainly, the technology should be in this list as everybody uses containers for R&D, research or even in production. AWS Fargate and Google Anthos are the leaders here. But there is a strong countertrend of running kubernetes and containers in VMs; the split here is about 30% for cloud native services and the rest for on-premise (I mean VMs but not private clouds, of course).
Other takeaways are:
- There is no industry trend for ML: there are more than 10 sets of technologies used with more or less the same percentage. I expected Sagemaker to be a leader but it has only 12%.
- Object storage is used by 49% of companies, but the majority of them struggle to clean up resources there.
- More than 40% of companies provision and manage clouds via scripts like terraform, chef and puppet.