The collective: a cache-based system management architecture

  • Authors:

  • Ramesh Chandra;Nickolai Zeldovich;Constantine Sapuntzakis;Monica S. Lam

  • Affiliations:

  • Computer Science Department, Stanford University, Stanford, CA;Computer Science Department, Stanford University, Stanford, CA;Computer Science Department, Stanford University, Stanford, CA;Computer Science Department, Stanford University, Stanford, CA

  • Venue:
  • NSDI'05 Proceedings of the 2nd conference on Symposium on Networked Systems Design & Implementation - Volume 2
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper presents the Collective, a system that delivers managed desktops to personal computer (PC) users. System administrators are responsible for the creation and maintenance of the desktop environments, or virtual appliances, which include the operating system and all installed applications. PCs run client software, called the virtual appliance transceiver, that caches and runs the latest copies of appliances locally and continuously backs up changes to user data to a network repository. This model provides the advantages of central management, such as better security and lower cost of management, while leveraging the cost-effectiveness of commodity PCs. With a straightforward design, this model provides a comprehensive suite of important system functions including machine lockdown, system updates, error recovery, backups, and support for mobility. These functions are made available to all desktop environments that run on the x86 architecture, while remaining protected from the environments and their many vulnerabilities. The model is suitable for managing computers on a LAN, WAN with broadband, or even computers occasionally disconnected from the network like a laptop. Users can access their desktops from any Collective client; they can also carry a bootable drive that converts a PC into a client; finally, they can use a remote display client from a browser to access their desktop running on a remote server. We have developed a prototype of the Collective system and have used it for almost a year. We have found the system helpful in simplifying the management of our desktops while imposing little performance overhead.