Reinventing ordering and locking for your distributed system
Modern, distributed systems are expected to not be bottlenecked by geographical distances, hardware- or software-failures or misconfigurations. Due to this, we did away with the mutexes, semaphores and global ordering we maintained in our monoliths. Some (mostly academic) systems reintroduced these features, usually through consensus-groups, which made the solution somewhat less centralised, but still exposed to the failure of a central group of these computers, where they are either closely located to each other or their throughput suffers.
These mutexes are also implemented synchronously, so one process has to wait for the other to finish by blocking on its thread, which doesn’t sit well with event-driven architectures.
We would like a solution which gives us the following:
- a pluggable service which lives with the microservice module
- is decentralised, no requirement for geographic proximity
- is asynchronous, and (ideally) also optimistic
- can run on both cloud and on-prem computers
Ours is the first such technology in the world. We achieve all this by rethinking the consistency model from scratch, to provide a solution which is unmatched by anyone else. We can retrofit this solution on your existing platform and allow your engineers to focus on new features instead of the endless slog of concurrency problems.
Take a look at how we achieved this, or reach out to us via our website.