Ever wonder what happens when you click on my.cwu.edu?
When your web browser connects to a major site like www.amazon.com, it is not accessing a single physical server; rather, it connects to a cluster of front end devices which intelligently directs traffic to groups of virtual back end web servers, which in turn retrieve content from database servers and other sources and return it to your browser. Such a configuration distributes the load across multiple servers, provides redundancy in case of server failure and enhances security by isolating back end systems.
While much smaller in scale, the architecture of MyCWU is similar. When your browser connects to my.cwu.edu, it actually connects to a virtual IP address (VIP) on a device called a load balancer. Behind the load balancer, there are multiple virtual portal web servers running under VMware, and the balancer directs your browser's traffic to one of these servers and ensures that you always connect to the same one throughout your session.
The portal web server runs two web server processes. The first is Apache, which intercepts your request and redirects you to the CWU single sign on page. Apache then proxies your request to the PeopleSoft web server. The PeopleSoft web server communicates with another virtual server called an application server, which in turn communicates with a virtual database server where most of the content is stored. This content is returned to your browser via the same path in reverse. Thus we have a multi-tiered architecture in which traffic flows as follows:
browser -> my.cwu.edu (load balancer) -> portal web server -> application server -> portal database server -> application server -> portal web server -> load balancer -> browser
It is actually a bit more complicated than this. MyCWU is a portal, and most content is provided by systems other than the portal web server. As you navigate around MyCWU, perhaps you have noticed URLs containing my-hrprd.ea.cwu.edu, my-csprd.ea.cwu.edu or my-fsprd.ea.cwu.edu. These are the VIPs for the content providers for the human resources, student and financial systems. Depending on which link you click while logged in to MyCWU, your browser may communicate directly with these content provider VIPs, or the portal may function as a web client itself and retrieve content from the providers on your behalf. In the first case, the data flow is similar to the one above. In the second case, however, the data flow is much more convoluted. E.g., an individual content frame (gadget) on the main portal page which displays employee information depends on a data flow similar to the following:
browser -> my.cwu.edu (load balancer) -> portal web server -> my-hrprd.ea.cwu.edu (load balancer) -> content web server -> application server -> HR database server -> application server -> content web server -> load balancer -> portal web server -> load balancer -> browser
Gadgets which retrieve data from the student and financial systems operate similarly, but interact with different servers. In all, there are more than 40 virtual servers working behind the scenes to provide the functionality of MyCWU. These servers reside in a virtual hosting environment consisting of multiple blade chassis and a large networked storage system. In addition, numerous network devices such as routers, switches and firewalls are involved.
Complex? Indeed. A lot happens when you click your mouse.