This chapter is from the book
This chapter is from the book
This chapter is from the book
Network Topology
Network topology refers to your network's layout, or how you link its components together. Network topology determines hardware links and how data flows across them, and thus has security implications.
When choosing a topology, consider these risks:
The single point of failure—A central point (a hub, wire, router, switch) on which one or more network devices rely. When this central point fails, the system can lose network connectivity, and your site will be down. Every network has one single point of failure, and some have more than one. Your aim is to minimize the damage a network outage can cause, and different topologies pose different limitations in this regard.
Susceptibility to electronic eavesdropping—Electronic eavesdropping is where attackers surreptitiously capture network traffic. All topologies are vulnerable, but some topologies offer greater security than others.
Fault tolerance—In this context, this is your network's capability to survive isolated failures. That is, if one, two, or five workstations fail, will remaining workstations continue to operate? If your network is fault tolerant, the answer is yes.
Unless you have reasons not to, choose star topology, and implement it with hubs, switches, or routers that support encryption, access passwords, and administrative authentication. Also, run your wire through the walls, instead of exposing it where others can physically access it. Finally, reduce your Web system's complexity whenever possible.
NOTE
For a good, quick primer on what various topologies look like, go to http://fcit.coedu.usf.edu/network/chap5/chap5.htm.
For example, don't distribute functions on a machine-by-machine basis unless you must. You've probably seen this before: one machine stores images, another stores CGI, another stores bare content, denial-of-service isn't necessary to discourage visitors—partial denial-of-service can, too.
Suppose that your developers build dynamic pages with media and logic housed on many different machines. What happens if one of those machines dies? You've seen this when a page never paints because it's waiting for images from other servers, or it's trying to send a transactional log elsewhere, to another network. Users have no patience, and if your site offers commerce services, these failures can cost you dearly. Systems parted out in the aforementioned manner are more likely to become partially disabled by malicious actors.