Crash Only Software - HotOS IX — Paper
https://www.usenix.org/legacy/events/hotos03/tech/full_papers/candea/candea_html/index.htmlCrash only software is more or less as the name suggests - there is no "normal" shut down procedure, the program just dies. This can lead to improvements in robustness with less code because: - you always need to handle the failure case anyway, so you may as well do it well. If you can always safely restart you don't need to worry about graceful shutdow. - even in the best of circumstances, your code is going to crash, or die when the machine it's running on does, so you may as well handle it now - all startup situations for software are uniformly about recovery - there's always a consistent process to follow - also talks about stateless applications as a key idea, with persistence of state kept to separate dedicated systems (oh hi there stateless web serving layers)
Tags
Related By Tags
- 🔗 The Rise of ``Worse is Better''
- 🔗 Why are we so bad at software engineering? | www.bitlog.com
- 🔗 Dan McKinley :: Choose Boring Technology
- 🔗 Performance is a UX problem | Clearleft
- 🔗 Being a responsible subsystem – Jessitron
- 🔗 Game UI Database | Welcome
- 🔗 InvisibleUp - Articles - FrontPage: The Good, The Bad, and The Ugly
- 🔗 10 Years of Open-Source Visualization / Mike Bostock / Observable
- 🔗 A parable about problem solving in software development | David R. MacIver
- 🔗 The C4 Model for Software Architecture
Details
- Revised
- Created
- Edited