Web Site Infrastructure Considerations
Guest editor: Wes Chou, Cisco
Total pages: 43
$29.00
Introduction
Virtually all Web-centric data centers follow the three-tier model: Web sever
tier, application server tier, and database server tier. While the specific
topology in which these tiers are configured depends on the service provided,
the fundamental infrastructure components are common across deployments. The
method of load balancing traffic to the sites, the means of replicating data
throughout the site, and even the physical layout of the servers are all issues
to consider in data center design.
This collection of articles touches on the infrastructure components, providing
an overview of the typical data center and setting the foundation for
intelligent tradeoffs when analyzing a Web site�s specific scaling and
application needs.
The first article, "Architecture and Dependability of Large-Scale Internet
Services," provides an excellent overview of popular Web site data centers, and
characterizes several distinct deployment models. Not only are the key
components of the data center highlighted, but the reasoning behind the
configuration choices is explained. The next article, "Lessons from Giant-Scale
Services," goes into detail about the actual vendor and design options, focusing
more on the load balancing aspects than the database aspect. Although the
article is slightly dated in terms of the actual vendor offerings (it seems that
vendors are coming out with newer models every year, or every other year), the
concepts remain relevant in today�s data center architectures.
Focusing slightly more on the database tier, "Data Management Issues in
Supporting Large-Scale Web Services," provides a first-hand look at some of the
design decision and data center components involved in storing and replicating
data for a major Internet application (Hotmail).
Rounding out this collection, "Power and Energy Management for Server Systems,"
analyzes the power consumption of both Web and database servers, describing the
various usage levels depending on the type of work they are doing. Although this
aspect of data center design is less technical, it remains an important aspect
when focusing on green technology, as well as the minimizing of operational
costs.
With this collection, readers will walk away with an idea of the core
infrastructure involved in Web site data centers design.
Keywords: large-scale Internet services, Web services, multitier architecture,
dependability, replication, graceful degradation, disaster tolerance
Table of Contents
Architecture and Dependability of Large-Scale Internet Services
David Oppenheimer and David A. Patterson, University of California at Berkeley
An analysis of the architectures and causes of failure at three large-scale
Internet services can help developers plan reliable systems offering maximum
availability.
Lessons from Giant-Scale Services
Author names and affiliations: Eric A. Brewer, University of California,
Berkeley
Giant Web services require new tools and methods for issues of scale,
availability, and evolution.
Data Management Issues in Supporting Large-Scale Web Services
Philip A. Bernstein, Nishant Dani, Badriddine Khessib, Ramesh Manne, and David
Shutt, Microsoft
This article discusses technical problems that arise in supporting large-scale
24�7 Web services. Issues covered include multitier architecture, costs of
commodity vs. premium servers, managing replicas, managing sessions, use of
materialized views, and controlling checkpointing.
Power and Energy Management for Server Systems
Ricardo Bianchini, Rutgers University
Ram Rajamony, IBM Austin Research Lab
This survey shows that management techniques tailored to different types of
servers and their associated workloads can provide substantial energy savings
with little or no performance degradation.
Recommended Resources
