Keywords: Transactions, Real-time order, HTAP, OLDA
TL;DR: In this paper, we design SnapKV, a key-value store that supports heterogeneous workloads without sacrificing real-time order.
Abstract: Modern businesses increasingly require fast and accurate real-time decisions over concurrently modified data. The existing literature lacks a unified data repository supporting heterogeneous workloads (transactional, analytics, and machine learning inference). Prior work made significant trade-offs between freshness, performance, consistency, and generality. In this paper, we design SnapKV, a key-value store that supports heterogeneous workloads without sacrificing real-time order.
Area: Active Storage
Type: Solution
Conflicts: All(Lehigh University),Lewis Tseng(UMass Lowell),Binoy Ravindran (VT)
Potential Reviewers: Antonio Barbalace,Sudarsun Kannan
Revision: Yes
Previous Version: https://openreview.net/forum?id=cEKUEWOfdG#discussion
Contact Email: dsm220@lehigh.edu
Submission Number: 3
Loading