Affinity-aware Serverless Function Scheduling (2407.14572v1)

Abstract: Functions-as-a-Service (FaaS) is a Serverless Cloud paradigm where a platform manages the scheduling (e.g., resource allocation, runtime environments) of stateless functions. Recent developments show the benefits of using domain-specific languages to express per-function policies, e.g., policies can enforce the allocation of functions on nodes that enjoy lower data-access latencies thanks to proximity and connection pooling. Here, we focus on affinity-aware scenarios, i.e., where, for performance and functional requirements, the allocation of a function depends on the presence/absence of other functions on nodes. We first present aAPP, an affinity-aware extension of a declarative, platform-agnostic language for defining custom function scheduling policies. We implement a prototype supporting this scheduling language by extending the popular Apache OpenWhisk FaaS platform and show that using aAPP in affinity-aware scenarios leads to an appreciable reduction in latency without noticeable overhead for scenarios without affinity constraints.