22 april 2020Blogg

One year of cloud

It has been more than 12 months since I took a leap into the exciting world of cloud consulting. Now it is time to look back and throw some predictions for the rest of 2020.

I started my career with cloud at Hewlett Packard aeons ago. At that time the cloud was really immature and we had a huge variety of problems just on basic IaaS deployments. However, even then the benefits were so imminent that we thought it was really worth it. Customers were eager to have faster deployments, so somewhat unstable platforms did not seem like a big problem at all.

Along came OpenStack with its promise to simplify and unify things. I still have nightmares about upgrading OpenStack installations (Although I have heard that it has gotten better now). Yes, we did some NFV -related things with it as it was the only option back then.

The world has changed a lot during these years and in the last 12 months I think the pace is speeding up even more.

Joining Solita

I joined the Solita cloud consulting unit last January with high hopes and a little bit scared. But after the warm welcome I got from my colleagues and a comprehensive introduction to working at Solita, I immediately felt at ease. This would be the place where I can make a real impact. Working for Solita has proven to be something that I imagined it would be. Lots of freedom, but also lots of responsibility to take. This is not for everyone, but for me it works.

During these months I have seen that the demand for cloud competency has risen into a whole new level. Customers are really looking to get serious with cloud adoption, if they have not yet done so. Microservice architecture with containers and FaaS based offering is really taking off even more.

One year of cloud blog

My projects have varied a lot during this year, ranging from simple cloud deployments to complete cloud strategy and governance projects.

One of the key skills here is that you need to put yourself into the shoes of the client and you really need to be aware of all aspects of modern business. Only that way you really can evaluate how to use the cloud for the clients best interests. I wont go into details in this posts on what has happened on the tech side during these 12 months, as the cloud evolves so fast that it is pointless to list all new features.

Crystal ball

I estimate that there is some demand for on-premises cloud during 2020. Azure stack and AWS outpost both seem to be pretty interesting and I hope we see some cases with those. The price tag is quite big, but not impossible, if you have a real use for them.

Serverless will keep on winning as the benefits with it are so obvious, even if there are some hiccups with the management layer from time to time (this needs its own blog entry).

It’s difficult to make predictions, especially about the future. But I will make one clear prediction that FaaS will rule even more during 2020. There are some clear benefits with especially DevOps minded developers. It will take the focus away from infrastructure even more and true IaC deployments are easier to do. It is more cost effective as resources are only used when code is executed. Scaling is easier and more imminent than on legacy VM’s or containers.

There are some rules of thumb to keep in mind when working with FaaS services. Limit single FaaS to do a single action, limit scope and keep functions lightweight. Pay attention when using libraries, they have a tendency to slow down functions. Keep functions isolated, don’t call functions from other functions.

Cold starts

With Faas there is the inevitable discussion about cold starts. The problem is that when running Functions which are not “warm”, it takes some time to bring the infra up and running. Usually the delay is around 2-3 seconds which is a deal breaker for some usage. But there are some workarounds for this and vendors are constantly improving this.

Also AWS is a bit ahead here and announced provisioned concurrency at 2019 Re:invent which basically keeps your lambdas “warm” and cuts the latency. Look for a summary here.

Check out Mikhail Shilkov analysis on coldstarts with AWS, Azure and GCP

To summarise last year: It has been a very busy year for our team and I think in 2020 we are going to do some great things again.