Distributed Programming via Safe Closure Passing

Programming systems incorporating aspects of functional programming, e.g., higher-order func- tions, are becoming increasingly popular for large-scale distributed programming. New frameworks such as Apache Spark leverage functional techniques to provide high-level, declarative APIs for in- memory data analytics, often outperforming traditional “big data” frameworks like Hadoop MapRe- duce. However, widely-used programming models remain rather ad-hoc; aspects such as implemen- tation trade-offs, static typing, and semantics are not yet well-understood. We present a new asyn- chronous programming model that has at its core several principles facilitating functional processing of distributed data. The emphasis of our model is on simplicity, performance, and expressiveness. The primary means of communication is by passing functions (closures) to distributed, immutable data. To ensure safe and efficient distribution of closures, our model leverages both syntactic and type-based restrictions. We report on a prototype implementation in Scala. Finally, we present pre- liminary experimental results evaluating the performance impact of a static, type-based optimization of serialization.

Instant Pickles: Generating Object-Oriented Pickler Combinators for Fast and Extensible Serialization

Eugene Burmako, Philipp Haller, Martin Odersky

As more applications migrate to the cloud, and as “big data” edges into even more production environments, the performance and simplicity of exchanging data between compute nodes/devices is increasing in importance. An issue central to distributed programming, yet often under-considered, is serialization or pickling, i.e., persisting runtime objects by converting them into a binary or text representation. Pickler combinators are a popular approach from functional programming; their composability alleviates some of the tedium of writing pickling code by hand, but they don’t translate well to object-oriented programming due to qualities like open class hierarchies and subtyping polymorphism. Furthermore, both functional pickler combinators and popular, Java-based serialization frameworks tend to be tied to a specific pickle format, leaving programmers with no choice of how their data is persisted. In this paper, we present object-oriented pickler combinators and a framework for generating them at compile-time, called scala/pickling, designed to be the default serialization mechanism of the Scala programming language. The static generation of OO picklers enables significant performance improvements, outperforming Java and Kryo in most of our benchmarks. In addition to high performance and the need for little to no boilerplate, our framework is extensible: using the type class pattern, users can provide both (1) custom, easily interchangeable pickle formats and (2) custom picklers, to override the default behavior of the pickling framework. In benchmarks, we compare scala/pickling with other popular industrial frameworks, and present results on time, memory usage, and size when pickling/unpickling a number of data types used in real-world, large-scale distributed applications and frameworks.

2013

Instant Pickles: Generating Object-Oriented Pickler Combinators for Fast and Extensible Serialization

Eugene Burmako, Philipp Haller, Martin Odersky

As more applications migrate to the cloud, and as "big data" edges into even more production environments, the performance and simplicity of exchanging data between compute nodes/devices is increasing in importance. An issue central to distributed programming, yet often under-considered, is serialization or pickling, i.e., persisting runtime objects by converting them into a binary or text representation. Pickler combinators are a popular approach from functional programming; their composability alleviates some of the tedium of writing pickling code by hand, but they don't translate well to object-oriented programming due to qualities like open class hierarchies and subtyping polymorphism. Furthermore, both functional pickler combinators and popular, Java-based serialization frameworks tend to be tied to a specific pickle format, leaving programmers with no choice of how their data is persisted. In this paper, we present object-oriented pickler combinators and a framework for generating them at compile-time, called scala/pickling, designed to be the default serialization mechanism of the Scala programming language. The static generation of OO picklers enables significant performance improvements, outperforming Java and Kryo in most of our benchmarks. In addition to high performance and the need for little to no boilerplate, our framework is extensible: using the type class pattern, users can provide both (1) custom, easily interchangeable pickle formats and (2) custom picklers, to override the default behavior of the pickling framework. In benchmarks, we compare scala/pickling with other popular industrial frameworks, and present results on time, memory usage, and size when pickling/unpickling a number of data types used in real-world, large-scale distributed applications and frameworks.

Assoc Computing Machinery2013

Spores: A Type-Based Foundation for Closures in the Age of Concurrency and Distribution

Philipp Haller, Martin Odersky

Functional programming (FP) is regularly touted as the way forward for bringing parallel, concurrent, and distributed programming to the mainstream. The popularity of the rationale behind this viewpoint (immutable data transformed by function application) has even lead to a number of object-oriented (OO) programming languages adopting functional features such as lambdas (functions) and thereby function closures. However, despite this established viewpoint of FP as an enabler, reliably distributing function closures over a network, or using them in concurrent environments nonetheless remains a challenge across FP and OO languages. This paper takes a step towards more principled distributed and concurrent programming by introducing a new closure-like abstraction and type system, called spores, that can guarantee closures to be serializable, thread-safe, or even have general, custom user-defined properties. Crucially, our system is based on the principle of encoding type information corresponding to captured variables in the type of a spore. We prove our type system sound, implement our approach for Scala, evaluate its practicality through an small empirical study, and show the power of these guarantees through a case analysis of real-world distributed and concurrent frameworks that this safe foundation for migratable closures facilitates.

2014