Novanet på NDC Oslo 2020

Going fast and cheap with Bots, Sanity and Kubernetes

Sometimes we need to go fast and we may find ourselves on a tight budget. In these situations it is important that we spend our time and resources right. It is still possible to be professional and apply best practices.

In this talk Olav will present lessons learnt from a hobby project building a system for organizing players in an MMO guild. This will include:
- Using bots in free chat apps for rapid development of user interactions,
- Using the chat app for OAuth authentication
- Sanity for both free headless CMS, database and automatically generated management UI
- Secure and scalable backend on secretless Kubernetes using Aks, Azure pod identity and keyvault

Se innlegget til Olav her

When each millisecond counts

When users make millions request to your site every single performance bottleneck could present a serious problem. Every millisecond becomes critical for your service stability. Where the line between premature optimization and neccessary adjustments should be drawn? This session is intended to share and discuss common mistakes and antipatterns causing performance degradation. There will be given examples on how to achieve better performance both from WebApi prospective (.Net) and the database layer. We will also discuss how to test and estimate your service performance with k6 before going live to production. The talk is intended for backend developers primarily and is based on work done for - norwegian weather forecast service used worldwide with more than 10 millions unique users per week.

Se innlegget til Dmitry her

Serving election results to an entire nation

What do you have to consider when all of Norway will be accessing your web site at exactly 9 PM, and you get only one shot to deliver?
During 2019 Hallstein worked in a two-man team for NRK, building a greenfield .NET Core API to serve election results to TV, web and more. The election was the most important project for NRK during 2019, and it was absolutely crucial that everything worked as planned. Hallstein will give you insights on how they approached this task, with an absolute deadline and inescapable performance demands, and how they flawlessly coped with one of the highest peak traffic loads in Norway during 2019 when other well known media sites broke down.

Se innlegget til Hallstein her

Breaking the build with K6 load tests

Performant services are increasingly important, and every change to your service can affect that performance. So, what if you could make load tests part of your release pipeline? In this talk we'll look at K6, an open source command line load testing tool, and show how you can check if your APIs can handle a massive burst of users in a short period of time. We'll also look at how you can add load tests to your Azure Devops Pipeline and make the build fail if the APIs don't perform.

Se innlegget til Lars Alexander her