Formal foundations of serverless computing
Formal foundations of serverless computing Jangda et al., OOPSLA’19
Jangda et al. won a distinguished paper award at OOPSLA this year for their work on ‘Formal foundations of serverless computing.’ Central to the paper is their observation that the serverless execution environment has a number of unique properties (such as warm starting / reuse of execution environments) that can make it harder to build correct applications. They show the conditions under which a serverless function can safely ignore these peculiarities, and thus become much simpler to reason about. They also introduce a composition language based on arrows for function composition.
Things to be aware of when going serverless
The serverless computing abstraction, despite its many advantages, exposes several low-level operational details that make it hard for programmers to write and reason about their code. For example, to reduce latency, serverless platforms try to reuse the same function instance to process multiple requests. However, this behaviour is not transparent and it is easy to write a serverless function that produces incorrect results or leaks confidential data when reused. A related problem is that serverless platforms abruptly terminate function instances when they are idle…
These first problems are related to Continue reading
