AHA Moments from the Public Cloud Meetup
At the end of August, we organized a #HeurekaDevsMeetup covering the public cloud, partially because Heureka Group is testing a hybrid model across three countries and we run into pitfalls that someone definitely encountered with before we did. One of the main reasons we organized the meetup was to discuss authentic use-cases with other experts and independently exchange experience across companies and sectors. This sharing format allowed us to break up into small groups and discuss the technological and managerial problems that come with migrating to a public cloud. This is where we found our Aha moments, and we hope they will serve as inspiration for further discussions for you as well.
We Are All Dealing with the Same Stuff
- Those who are starting with the cloud are afraid they are behind others who are already smoothly operating in the cloud. On the contrary, many companies are still trying to figure out how to manage.
- The tech community in Prague may be small, but we don’t talk to each other enough. That’s why we’re all dealing with the same things… over and over again.
What to Take into Account in Decision Making?
- Sometimes it was necessary to decide on the basis of corporate strategy or policy: The decision did not always come from developers or the infrastructure team.
- Going into the cloud or not isn’t the question. When and what to choose are more important. Calculating the on-prem costs and learning billing can be done anytime.
- Infracost – check out this service that helps estimate cloud operational costs.
- As an example: It’s possible to show internally what percentage each team costs from the total cost of the product. The teams can then challenge each other as no one wants to be the most expensive. However, this can also have a negative effect. Teams might not want to add new functionalities because it would increase costs.
- Geolocation and latency for customers
- It’s not necessary to be bound to the datacentre that is the closest (i. e. Frankfurt for GCP). We won’t find anything closer for Hungary and Slovenia where we have other infra teams.
- There is a fear of higher latency, but edge points of presence (PoP) mitigate these fears. Discussions about the infra team after migrating to the cloud
A Cultural Fit is a Necessity
- We heard: „If I had to do it again, I would first spend six months educating developers and then begin with migration.“
- The cloud isn’t just about technologies, but mostly about a mindset, because people must change their behaviour, habits, and methods of working. That is why you need doers = people who by themselves want to work in the cloud. It’s not just about capabilities. But careful, not everyone wants to change or understand that migration is a good thing and new capabilities are necessary.
- Count on losing some people, unfortunately.
DevOps and Maintenance
- What happens to the infra team after configuring the cloud?
- The foundations can be built quickly, but you can always go a step further. They can allow you to do what you like and then move forward.
- There will always be something to do. People will work as „internal consultants“ to developers, helping them to provide better services and transfer knowledge to facilitate growth.
- Sometimes current developers can learn the DevOps approach. Again, a cultural fit is important for changes to take place.
- Cloud support is a problem:
- There aren’t many companies that would be large enough to create better and faster support.
- We also heard that being a larger client in a smaller public cloud can mean faster reactions to support requests and incorporating changes.
- Heard at the meet-up: „We all know hardware is a fetish.“ The various topics we discussed at the meetup.
Hybrid Whether We Want It or Not
- Our strategy
- We don’t move everything from on-premises to the public cloud (it would be prohibitively expensive)
- We take new services/applications into the cloud as part of unifying Heureka Group platforms, and for this we have a hybrid cloud.
- We confirmed in the end that no one is completely in the cloud. Each model eventually becomes a hybrid whether it is desired or not. Some „purely cloud“ companies have their CDNs on local, for example.
- Some use on-premises for certain use cases that would otherwise be too costly.