In-Service Performance Monitoring
Once a network circuit has been activated, and during subsequent 'in-service' usage, the challenge for the Service Provider becomes one of monitoring actual performance of the network under 'real load' conditions. For pure Layer-2 (Ethernet) networks, the ITU-T Y.1731 protocol provides mechanisms for enabling periodic test packets to be inserted into live data streams, which can then be used to monitor latency and jitter characteristics for each service defined across a Layer-2 circuit. The size and frequency of such test packets are low, hence they are 'minimally invasive' and do not appreciably impact the performance or characteristics of customers' services.
Recently, the more advanced Demarcation Devices, such as those within our MetroCONNECT EDD family, have started to include an additional protocol suite, the Two-Way Active Measurement protocol' (TWAMP), which performs a similar job, but for IP-addressed traffic. TWAMP is also somewhat less complex in set-up and operation than Y.1731, so is seen both as more simple to use and more flexible, since it can be used just as readily for layer-2 networks as it can for networks comprising partial or complete Layer-3 (IP Routed) infrastructure.
In operation, for one or more distinct services or 'flows' defined for customer traffic within a Carrier's network, an 'SLA profile' may be established between pairs or numbers of EDDs at different customer locations. Such profile may include the following:
- Service throughput
- Maximum allowable latency
- Maximum allowable delay variation/jitter
- Maximum allowable traffic frame loss ratio
- Minimum allowable 'availability' percentage
Against each of these parameters, it is possible to assign alarm criteria such that, for example, the Carrier's Network Operations Centre might be alerted if the latency of one of a customer's services, perhaps representing only a single virtual pathway within an overall Ethernet 'pipe', rises above a certain threshold defined within the customer's Service Level Agreement.
Whilst alarming in the event of specific threshold criteria being breached is clearly important, a Service Provider may wish alternatively to view trends in key performance criteria, for example the variation of traffic delay during a one day or one month period.
Our MetroCONNECT family of Demarcation Devices offer great flexibility in the viewing and retrieval of such data. in the first instance, real-time characteristics are stored in a format which can be readily accessed by SNMP-based network Management Applications. Management console operatives can readily set up an enquiry to interrogate key data on a regular basis, say every 15 minutes, thereby building up a table, or graph, of any service performance variable which may be of concern. Alternatively, the products themselves store an appreciable amount of historical data, filed within 15-minute, 1 hour and 24 hour logs for up to 30 days, available to view dynamically by log-in under management control to any Demarcation Device.
In-service Performance Monitoring
TWAMP 'background' monitoring can be used to generate alarms to the Network Operations Centre in the event that (Latency, Jitter, Packet-Loss) becomes out-of-specification and breaches pre defined 'Service Level Agreements' for each link.
Operatives at the NOC can remote log-in to a unit to:
- Configure SLA thresholds
- Initiation of background monitoring
- SNMP 'Trap' alarm reception (SLA breach)
- Further investigation by individual unit log-in
Network Performance, management and provisioning
Networking The World Since 1989