# Slack Squeeze Coded Computing for Adaptive Straggler Mitigation

## TL;DR

This paper introduces Slack Squeeze Coded Computation ($S^2C^2$), a dynamic workload distribution method that reduces latency in distributed computing by efficiently managing compute slack based on node speed predictions.

## Contribution

The paper proposes a novel adaptive workload distribution strategy for coded computation that minimizes overhead without redistributing data, using an LSTM-based speed prediction.

## Key findings

- Achieves 19% to 39% latency reduction over traditional methods
- Effective across various algorithms including linear algebra and graph processing
- Demonstrates broad applicability beyond matrix-vector multiplication

## Abstract

While performing distributed computations in today's cloud-based platforms, execution speed variations among compute nodes can significantly reduce the performance and create bottlenecks like stragglers. Coded computation techniques leverage coding theory to inject computational redundancy and mitigate stragglers in distributed computations. In this paper, we propose a dynamic workload distribution strategy for coded computation called Slack Squeeze Coded Computation ($S^2C^2$). $S^2C^2$ squeezes the compute slack (i.e., overhead) that is built into the coded computing frameworks by efficiently assigning work for all fast and slow nodes according to their speeds and without needing to re-distribute data. We implement an LSTM-based speed prediction algorithm to predict speeds of compute nodes. We evaluate $S^2C^2$ on linear algebraic algorithms, gradient descent, graph ranking, and graph filtering algorithms. We demonstrate 19% to 39% reduction in total computation latency using $S^2C^2$ compared to job replication and coded computation. We further show how $S^2C^2$ can be applied beyond matrix-vector multiplication.