The scale of public software platforms – We can do it better!
Is the COVID-19 test platform “Österreich testet” really worth more than €500,000, or is a clone that was created within 1 hour capable of providing the same service? Is it really a good PR strategy to argue that approximately 5000 applications per hour took down the platform of the COVID-19 support fund of the WKO? Why is the website of the Austrian Parliament not able to deal with the load that was generated by the assessment period of a new law?
Well, from a software architectural perspective, the answer always starts with “It depends.”
It depends on so many factors to assess what really is going wrong with these software systems. It would be unfair to judge them based on the information that is available to the public. What bothers me about these episodes is the poor light cast on the software industry in Austria.
How should anybody believe in the digitalization efforts of the Austrian government if they accept paying roughly €190,000 per month to run a rather simple software system? Politicians are no experts in software development, and they don’t run software companies on their own. So, it is a good idea to outsource development and rely on professionals. I don’t have the insights or intentions to doubt these professionals. Still, I would argue that any private company that runs such an expensive system to provide a limited feature set would be out of business soon.
The 1h long experiment conducted by 4 developers during the TV show “Willkommen Österreich” shows that their prototype implements the basic use-case. This version is surely not a complete replica of the original, but it could indicate that this platform could have been cheaper.
Further, as an industry, it looks like we are incapable of designing and building systems that can handle 5000 simple requests per hour. This is usually a scale that doesn’t and shouldn’t create headaches. While there might be a valid reason why the mentioned systems failed, using this number of requests as an excuse looks ridiculous. The Upper Austrian COVID-19 vaccination platform shows a different approach: they consulted professionals and can easily handle 5000 registrations per minute.
These episodes don’t reflect the Austrian software industry. I have worked with tons of motivated professionals and wonderful companies and startups that show daily that they can handle almost any scale. Solid software architectures are key to building reliable, resilient, and cost-effective software systems. It would be great if we could see more of this expertise leveraged by our government.
Media coverage of the mentioned episodes:
- Corona-Website “Österreich testet” kostete mehr als halbe Million Euro
- Vier Entwickler bauen umstrittene “Österreich testet”-Website in 50 Minuten nach
- Beim Härtefonds schauen einige Betriebe durch die Finger, Server teils überlastet
- Begutachtung zum “Freitesten”-Gesetz überlastet Webseite des Parlaments
- Erfolgsstory: Dynatrace optimiert Impfanmeldung für Land OÖ
Photo by Sarah Kilian