16.2. Algorithms
In Chapter 14 we considered a variety of points in the network where a request may be diverted to one of several server replicas. Any of these points has an opportunity to select the most appropriate server for the request. However, different metrics would be most appropriate or even obtainable at different points.
DNS-based request distribution (see Section 14.1.3) often uses a proximity metric as the primary server selection criterion. Using load metrics for fine-grained server selection is difficult at the DNS level because of the hidden load problem (see Section 15.5). Section 16.4 discusses DNS server selection in more detail.
On the other hand, server-load metrics are the primary metrics used by a load-balancing switch ...
Get Web Caching and Replication now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.
