One of the biggest headaches for a company’s network managers is verifying that contracted service levels are being met for their infrastructure. Carrier services usually include a series of contractual points which are summarized in service level agreements (SLAs). To determine whether a service meets expected quality standards, the customer and carrier will use the metrics stated in the SLAs as parameters for measurement.
To transfer this into practice, all end users need do when experiencing network problems (i.e., when the Internet isn’t working, connections with the central servers are slow, voice communications break up, etc.) is to open an incident triggering an alarm in the communications systems. The user reports a quality of experience (QoE) incident. These kinds of incidents have a number of things in common, namely, they are often subjective, sporadic and difficult to measure.
Quality of service (QoS) measures
Switches, routers, servers, etc., are equipped with built in analysis mechanisms that provide performance statistics and even report queue overflow problems, CPU usage, etc., caused by system anomalies. But how do we measure the state of the communications network that we have contracted?
To perform the analysis, we need realistic information that can reflect the problem that led the user to trigger an alert. And the way to do this is through contracted quality of service (QoS) measures. These measures are based on a series of metrics that provide objective data that can be easily analyzed and compared against incidents, without having to rely on the subjective end-user experience.
Typical QoS metrics include:
- Delay: length of time it takes for information to travel from source to destination. RTT – which is the length of time it takes for a data packet to make a round trip from source to destination and back- is often used as a measure.
- Packet loss: percentage of traffic that fails to reach its destination.
- Delay variation (Jitter): traffic flows may be delayed by the networks in different ways for different packets. This difference manifests itself markedly in multimedia traffic.
The service administrator can use these three values to get a basic idea of how a network is performing and even complete status reports. And if he has the right tools, he can add other types of metrics as well (e.g., bandwidth metrics).
Teldat and Naudit develop a new solution for measuring QoS
A practical application for the three aforementioned values might be that of a customer with n remote sites interconnected among themselves, or in a star topology with their head office, which might have a centralized measurement system showing the customer instantaneous values of these metrics and their accumulated values over time. Thus, an increase in user QoE incidents will likely cause an alteration of the normal QoS values of the network (and therefore the network administrator receives technical information backing a possible malfunction of the contracted service).
Teldat and its technology partner Naudit, have developed a joint QoS measurement solution of customer infrastructure. The solution consists of a measurement probe placed in each Teldat router at remote offices and at a centralized software providing the administrator via a control panel quality results and from where he can obtain and export statistical information about his network with the metrics discussed above.
Apart from being integrated into the routers, it is a highly precise measurement solution, which at the same time is not intrusive.
Together, Teldat and Naudit offer innovative technological solutions to complex problems.