Overview of Measurement and Tools
What specific measurements do we perform on NCREN, and what do they represent?
Measurements on NCREN can be categorized into three general areas:
1. General availability
Managed network devices, servers and services on NCREN are monitored for availability, typically by testing general network reachability via ICMP response measurement. For servers and services, service-specific tests are also issued as a measure of service availability. For both types of tests, measured results that exceed predefined thresholds trigger notification to NCREN Operations staff 24x7x365.
2. Traffic utilization measurement
NCREN utilizes and manages hundreds of network devices which provide connectivity services to all research and education institutions throughout the state. In order to ensure that network resources are being used effectively and as expected, utilization for all major circuits is measured and continuously monitored. Utilization measurements are stored for later review.
Measurements are typically viewed as graphs:
Traffic utilization is also monitored for threshold variance, and as with availability monitoring, exceptions trigger notification to NCREN Operations staff.
3. Latency / Loss / Jitter
While network performance can be impacted by a variety of factors, a common set of metrics for measuring relative performance is *latency* (delay between sending and receiving), *loss* (the relative amount of traffic that was not transmitted successfully), and *jitter* (variation in delay). Latency is usually measured in milliseconds (thousandths of a second), and ideally should be relatively low and consistent, with zero loss for optimal performance. Measured latency on the NCREN backbone is typically somewhere in the 2-8 millisecond range, with extremely small amounts of jitter and virtually zero loss. Rare exceptions in these measurements occur during unforeseen network events and outages; as with availability and utilization monitoring, exceptions trigger notification to NCREN Operations staff.
Graph of a typical connection showing low latency/loss/jitter: