Understanding The Components of a Streaming Media Network – Part 1

Load Balancing & Client Direction
With the content replicated out to the servers it becomes important to direct the user to the most "appropriate" streaming server. Appropriate can mean a lot of things varying from the least loaded (network, disk and CPU) server, to the closest server. Closes in network terms can be defined in a number of manners including geographic, network (IP subnet hops) and/or latency (ping times). The presence of the requested data may also factor into the calculations (i.e. it may be more expensive to send the request to an idle server which would have to "pull" and cache the server than to just send the request to a lightly loaded server that already has the content cached). The load balancing system may also incorporate an access control or load limiting functionality. This includes limiting access by IP address, aggregate egress bandwidth and/or concurrent viewers. Such limits are useful in preventing the network from becoming saturated. The network administrator may wish to globally limit the number of users and/or bandwidth being used across the entire network or as fine grained as limiting the number of concurrent viewers per server.

- How does the load balancing scheme measure server load?
- How does the client direction scheme determine the appropriate server?
- Does the load balancing scheme limit by user and/or egress bandwidth?

In Part Two
Part two of this article will cover the remaining components of a CDN infrastructure. This includes historical reporting, real-time monitoring, content management as well as the streaming/caching servers.

About JHL Consulting
JHL Consulting, founded in 1999, provides CDN (Content Distribution Network) engineering and consulting services. These services include network design and deployment as well as software development of digital rights, content management, reporting and load balancing systems for both public and private networks. www.JHLConsulting.com

About the Author
Mr. Studarus has been involved in streaming since 1995 as a graduate student working on the MBONE (Internet Multicast Backbone). He later developed a streaming media plugin for the Netscape browser predating both the Windows Media and RealPlayers. At InterVU (since purchased by Akamai), he developed software to run a public content distribution network used to broadcast many high profile events. With JHL Consulting, Mr. Studarus has engineered and deployed several public and private CDNs. This has included the software development and deployment of reporting, load balancing, digital rights management and content management systems. He holds a B.S. in Computer Science from the University of California, San Diego and a M.S. in Information Networking from Carnegie Mellon University.