The modern cloud presents developers with a fundamental dichotomy. At one extreme, Infrastructure-as-a-Service offers powerful but low-level building blocks that demand deep expertise in distributed systems to use safely. At the other extreme, cloud providers offer high-level managed services (SQL databases, key-value stores, managed versions of popular open-source systems) that are operationally simple but rigid, forcing applications to be built around whatever guarantees a given service happens to offer. What is missing is a middle layer: a set of composable, platform-agnostic abstractions that let developers assemble reliable, portable distributed applications by combining well-defined primitives for state, messaging, and computation, rather than by wiring together infrastructure or accepting the limits of an off-the-shelf system.

This missing middle cannot be designed from within a single community. The distributed systems community holds the foundational algorithms for consensus, replication, and fault tolerance, but these mechanisms are typically too intricate for direct use by application developers. The database systems community has decades of experience packaging declarative, developer-friendly semantics for state and consistency (SQL, ACID transactions) but almost exclusively within monolithic systems rather than as independent, reusable building blocks. The programming languages community brings the science of abstraction itself: type systems and formal methods capable of making composition safe and predictable. Progress on the missing middle requires these three perspectives to be brought together in the same room.

Our conviction is that this gap is not merely a matter of convenience but a structural bottleneck holding back an entire generation of cloud-native software. The goal of this Dagstuhl Seminar is to change the unit of reasoning for distributed application development, from assembling infrastructure to composing trustworthy abstractions. This shift is not a leap into the unknown: independently developed systems across the serverless, streaming, and transactional-database worlds keep re-deriving the same underlying structure, organizing state, communication, and computation as intertwined concerns rather than separate layers, since state is mutated in response to incoming messages, and computation is what drives both. This recurring pattern suggests that the abstraction we are after is already latent in practice – implicit in how these systems are built – and is waiting to be named, formalized, and studied as a first-class object, rather than rediscovered by every new system.

Concretely, we will work toward mapping the essential, reusable building blocks these systems are made of; envisioning the kind of runtime able to host and execute them portably, independent of the underlying cloud; and establishing the principles, grounded in formal specification rather than natural-language documentation, that let such building blocks be combined safely, so that guarantees proven for the parts (delivery semantics, atomicity, isolation) continue to hold for the whole.

Creative Commons BY 4.0

Elisa Gonzalez Boix, Asterios Katsifodimos, Alessandro Margara, and Guido Salvaneschi

This seminar qualifies for Dagstuhl's LZI Junior Researchers program. Schloss Dagstuhl wishes to enable the participation of junior scientists with a specialisation fitting for this Dagstuhl Seminar, even if they are not on the radar of the organizers. Applications by outstanding junior scientists are possible until Friday, October 9, 2026.