I expect that in time, the OSS community will abstract away these differences. We’ll get pluggable frameworks and libraries that will allow us to just provide configuration for the chosen cloud that we’re running on, and the framework will interface with the native services to make our platform-agnostic code run. We’re not there yet, and some of these services (especially in the AI space) are becoming so involved that it may never be possible to fully abstract them behind a common interface in a way that doesn’t critically compromise the functionality.
Coding for a serverless future
Jay Rodgers

That’s an important question that I also ponder. These services also evolve very quickly, much faster than OSS generalization efforts can go. There is a chance that by the time one wants to move to a new provider, the state of the art will have move so much that it may be advisable to rewrite for the new capabilities anyway. The only significant anchor that remains is a mass of data stored on one provider or another and the cost of moving this across.

Like what you read? Give Dmitry Serebrennikov a round of applause.

From a quick cheer to a standing ovation, clap to show how much you enjoyed this story.