Any backend components such as databases that the service depends may fail or impact 

performance 

How can performance problems be overcome? QoS technologies such as WebQoS and 

Diffserv can help solve application and network related performance issues [2, 4]. However 

for QoS guarantees to be possible, QoS needs to be implemented in all components in the 

service delivery chain. Clearly, there is little point in establishing network level QoS if the 

end application that is being accesses has no priority on which request to processes first. The 

problem for the Service Provider is that it does not manage all of the components in the 

service delivery chain so establishing appropriate QoS for all the components can be difficult 

between different organisations.  

In the context of the availability issues addressed above, redundancy may assist in making a 

service more reliable in that if one component fails in the delivery chain, another can take 

over to allow the service to continue. However, redundancy is usually not applied to all 

components in a service deliver delivery chain, if for any reason, the cost would be 

prohibitive. So if the customer's network connection to the Service Provider fails, the 

customer will not be able to access the service. However, the Service Provider may not be 

aware that the customers is experiencing difficulty accessing the service.  

Of course, guaranteed levels of QoS and availability of all the components gives no 

indication as to how the service performs from the end user perspective. Tackling this 

problem will be discussed next. 

3.4 Monitoring HTTP Delivered Services  

Having looked at how HTTP services are delivered, the next area to investigate is how can 

HTTP delivered services be monitored to give an accurate reflection of the service as 

perceived by the end user. Looking at types of Parameters to Monitor for Services Delivered 

by HTTP, they can be broken into two types: 

Performance parameters 

%