Dataframes are a popular abstraction to represent, prepare, and analyze data. Despite the remarkable success of dataframe libraries in R and Python, dataframes face performance issues even on moderately large datasets. Moreover, there is significant ambiguity regarding dataframe semantics. In this paper we lay out a vision and roadmap for scalable dataframe systems. To demonstrate the potential in this area, we report on our experience building Modin, a scaled-up implementation of the most widely-used and complex dataframe API today, Python's pandas. With pandas as a reference, we propose a simple data model and algebra for dataframes to ground discussion in the field. Given this foundation, we lay out an agenda of open research opportunities where the distinct features of dataframes will require extending the state of the art in many dimensions of data management. We discuss the implications of signature dataframe features including flexible schemas, ordering, row/column equivalence, and data/metadata fluidity, as well as the piecemeal, trial-and-error-based approach to interacting with dataframes.

Dataframes are a popular and convenient abstraction to represent, structure, clean, and analyze data during exploratory data analysis. Despite the success of dataframe libraries in R and Python (pandas), dataframes face performance issues even on moderately large datasets. In this paper, we take the first steps towards formally defining dataframes, characterizing their properties, and outlining a research agenda towards making dataframes more interactive at scale. We draw on tools and techniques from the database community, and describe ways they may be adapted to serve dataframe systems, as well as the new challenges therein. We also describe our current progress toward a scalable dataframe system, MODIN, which is already up to 30× faster than pandas in preliminary case studies, while enabling unmodified pandas code to run as-is. In its first 18 months, MODIN is already used by over 60 downstream projects, has over 250 forks, and 3,900 stars on GitHub, indicating the pressing need for pursuing this agenda.

Serving ML prediction pipelines spanning multiple models and hardware accelerators is a key challenge in production machine learning. Optimally configuring these pipelines to meet tight end-to-end latency goals is complicated by the interaction between model batch size, the choice of hardware accelerator, and variation in the query arrival process. In this paper we introduce InferLine, a system which provisions and manages the individual stages of prediction pipelines to meet end-to-end tail latency constraints while minimizing cost. InferLine consists of a low-frequency combinatorial planner and a high-frequency auto-scaling tuner. The low-frequency planner leverages stage-wise profiling, discrete event simulation, and constrained combinatorial search to automatically select hardware type, replication, and batching parameters for each stage in the pipeline. The high-frequency tuner uses network calculus to auto-scale each stage to meet tail latency goals in response to changes in the query arrival process. We demonstrate that InferLine outperforms existing approaches by up to 7.6x in cost while achieving up to 34.5x lower latency SLO miss rate on realistic workloads and generalizes across state-of-the-art model serving frameworks.

No abstract