A “web farm” is a high availability application architecture that is common to many vendors and products. It usually involves the use of multiple web (HTTP/S) application servers, each serving the same function, and often relying on the use of round-robin session distribution from a network load balancer (NLB). However, the term is also often applied to other servers that provide services to the web, notably FTP/S, SFTP and AS2 servers in the context of file transfer.
Web farms are used to provide horizontal scalability at a single location (e.g. adding additional web farm nodes in Dallas to expand capacity from 20K users to 40K users). They are also usually deployed in a multi-tier architecture, where data actually resides on “back end” database or file servers.
Web farms also fill a failover role in the sense that surviving web farm nodes can assume the duties of dead web farm nodes in an emergency. However, this is only true when the surviving web nodes have enough capacity to serve the remaining system requirements; lose too many web nodes and you lose your failover capability too.
BEST PRACTICES: If high performance is a requirement, managed file transfer solutions with web farm architectures are preferred. If deployed in Internet-facing servers, highly secure file transfer solutions should allow back-end data systems to reside on a separate “data zone” in a web farm configuration. Web farms at a single location should be able to be DR recovered at a second location through the use of SAN replication from location to location.