Clustering Concepts
The key to scaling a deployment is clustering. By grouping multiple servers together into a single cluster, you can leverage the resources of each server and distribute client workload across those servers. As site usage increases, you can add servers to the cluster to accommodate the increased workload.
Depending on site architecture and deployment design, clustering can also increase performance and availability by removing single points of failure and enhancing specific functionality. Isolating the deployment into separate clusters allows you to manage the clusters in each tier independently. In this manner, you can alleviate bottlenecks by isolating the problematic servers into their own cluster. For example, if processing transactions is the bottleneck, you can isolate this functionality onto its own database. You can then increase the availability and performance of this database through clustering, independent of the other servers in your deployment.
Typically, Commerce Server 2002 components use two key clustering methods for the following purposes:
- Network Load Balancing (NLB) on the Web tier to provide load balancing and scalability
- Cluster Service on the database tier to provide fault tolerance
Both clustering methods increase availability and reliability. Load balancing across the Web tier provides scalability and is more suitable for data that is stateless (such as data used for Web servers, streaming media, VPN, and proxy servers). Clustering the database tier is more suitable for data that is highly dynamic (such as data used for SQL Server, Exchange Server, file servers, and print servers).
A Web cluster using NLB typically consists of multiple cluster members. Each member runs as an independent Web server and does not share data with other cluster members (other than any data used for cluster operations). If any cluster member fails or encounters errors, this member can be set offline or taken out of the cluster so that incoming client requests are load balanced to working members only. For more information about NLB, see the documentation in Windows 2000 Advanced Server or Windows 2000 Datacenter Server.
A SQL Server database cluster using Cluster Service typically consists of two members. Both members share a common external storage device and are exact duplicates. If one member fails or encounters errors, the other member is substituted so that database functionality is operational. For more information about NLB, see the documentation in Windows 2000 Advanced Server or Windows 2000 Datacenter Server.
See Also
Copyright © 2005 Microsoft Corporation.
All rights reserved.