MTCache: Transparent Mid-Tier Database Caching in SQL Server

  • Authors:
  • Per-Åke Larson;Jonathan Goldstein;Jingren Zhou

  • Affiliations:
  • -;-;-

  • Venue:
  • ICDE '04 Proceedings of the 20th International Conference on Data Engineering
  • Year:
  • 2004

Quantified Score

Hi-index 0.00

Visualization

Abstract

Many applications today run in a multi-tier environmentwith browser-based clients, mid-tier (application)servers and a backend database server.Mid-tier databasecaching attempts to improve system throughput and scalabilityby offloading part of the database workload to intermediatedatabase servers that partially replicate datafrom the backend server.The fact that some queries areoffloaded to an intermediate server should be completelytransparent to applications - one of the key distinctionsbetween caching and replication.MTCache is a prototypemid-tier database caching solution for SQL Server thatachieves this transparency.It builds on SQL Server's supportfor materialized views, distributed queries and replication.This paper describes MTCache and reportsexperimental results on the TPC-W benchmark.The experimentsshow that a significant part of the query workloadcan be offloaded to cache servers, resulting in greatlyimproved scale-out on the read-dominated workloads ofthe benchmark.Replication overhead was small with anaverage replication delay of less than two seconds.