Orloj: Predictably Serving Unpredictable DNNs

Existing DNN serving solutions can provide tight latency SLOs while maintaining high throughput via careful scheduling of incoming requests, whose execution times are assumed to be highly predictable and data-independent. However, inference requests to emerging dynamic DNNs -- e.g., popular NLP models and computer vision (CV) models that skip layers -- are data-dependent. They exhibit poor performance when served using existing solutions because they experience large variance in request execution times depending on the input -- the longest request in a batch inflates the execution times of the smaller ones, causing SLO misses in the absence of careful batching. In this paper, we present Orloj, a dynamic DNN serving system, that captures this variance in dynamic DNNs using empirical distributions of expected request execution times, and then efficiently batches and schedules them without knowing a request's precise execution time. Orloj significantly outperforms state-of-the-art serving solutions for high variance dynamic DNN workloads by 51--80% in finish rate under tight SLO constraints, and over 100% under more relaxed SLO settings. For well-studied static DNN workloads, Orloj keeps comparable performance with the state-of-the-art.