Recently, I had a brief discussion on performance monitoring with a network engineer that works for an Internet Service Provider. After showing him a product tour of NetBeez, he mentioned that he could do the same with Ethernet performance tools (e.g. ITU-T Y.1731), which are available on most networking platforms, without the need of deploying additional hardware or software probes.
To support his statement, the network engineer mentioned that his requirements were to monitor individual Ethernet links, or portions of the overall network. I agreed with him that, for that specific use case, Ethernet performance monitoring was the right solution.
However, I reminded him that most enterprises and service providers that we work with use NetBeez to measure the end-user experience, and that Ethernet performance monitoring only runs at Layer 2. For this reason, the best way to monitor the network from the end-user perspective is to use Layer 3+ protocols that allow to run true end-to-end connectivity and performance tests against cloud applications, WiFi networks, and other services offered to users. This is not possible with Ethernet performance monitoring tools. If you want to read more about this topic, please check out this blog post that I wrote few weeks ago.
Performance Metrics
Ethernet performance monitoring has similarities with measurements available in NetBeez. In fact, the most common Ethernet performance metrics that are used by enterprises and service providers are:
- Frame Loss Ratio
- Frame Delay
- Frame Delay Variation (Jitter)
Each one of these parameters are key performance indicators that can be used to enforce Service Level Agreements with customers and applications. As you’ll read in the rest of this post, having a monitoring system in place that provides these values is extremely important.
Frame Loss Ratio
The Frame Loss Ratio (FLR) value is expressed as percentage, and is calculated by dividing the number of frames not received by the total number of frames sent. High Frame Loss Ratios can have negative impact on most applications, especially TCP based ones, where a retransmission causes the throughput of the connection to decrease.
In NetBeez, this value is represented by the packet loss value of ICMP- or TCP-based ping tests.
The Frame Loss Ratio can also be used to measure availability of a point-to-point link. In particular, the ITU states that when the FLR value is more than 30 percent for 10 consecutive measurement time slots, then the link is declared as down from the algorithm. On the other side, if the FLR drops below 30 percent for 10 consecutive time slots, the link is then declared available.
In NetBeez, by default, a resource is declared unavailable when five or more consecutive packets are not received by the source probe. This threshold can be manually changed by the user.
Frame Delay
The Frame Delay is one-way delay measurement between two endpoints. The calculation is made by comparing the timestamp included in the frame by the sender with the clock time when the frame is received. This parameter requires the clocks at both endpoints to be synchronized. If synchronization can’t be assured between the two endpoints, then only the two-way Frame Delay parameter can be calculated.
In NetBeez, the two-way Frame Delay value corresponds to the round-trip-time generated by ping tests.
Frame Delay Variation
The Frame Delay Variation is calculated by measuring the difference in delay recorded by different frames. Frame Delay Variation is also called jitter, and when this parameter has high values (> 100 ms), it will have an impact on the quality of real-time transmissions, such as video and voice calls.
In NetBeez, this value corresponds to the jitter value, which can be calculated with UDP-based iperf or VoIP tests. The VoIP tests also provide an estimate of mean opinion score.
Conclusion
As shown, you can see how Ethernet performance monitoring has very similar tests. This is not surprising because no matter what the application is, latency, packet loss, and jitter are very important key performance indicators of a network service. If you want to learn how you could monitor your network from the user perspective, request a product tour of NetBeez.
