Portal moves running Python workloads between machines without losing progress

Portal, a new infrastructure product, moves running Python workloads between machines while preserving their progress, which lets users run them on cheaper compute, according to a launch post on X. The post says the underlying principle was recently accepted to the ACM/IFIP Middleware conference.

Its author describes the approach as built at the application layer and traces earlier work on portable computing, from 1970s code portability through process migration and mobile agents. The report does not include independent evaluation of the product's performance or pricing.