Metrics (networking)

From Wikipedia, the free encyclopedia
Jump to: navigation, search
For other uses, see Metric (disambiguation).

Metrics is a property of a route in computer networking, consisting of any value used by a routing protocol to determine whether one particular route should be chosen over another. The routing table stores only the best possible routes, while link-state or topological databases may store all other information as well. For example, Routing Information Protocol uses hopcount (number of hops) to determine the best possible route. The route will go in the direction of the gateway with the lowest metric. The direction with the lowest metric can be a default gateway.

Examples[edit]

A Metric can include:

  • measuring link utilisation (using SNMP)
  • number of hops (hop count)
  • speed of the path
  • packet loss (router congestion/conditions)
  • latency (delay)
  • path reliability
  • path bandwidth
  • throughput [SNMP - query routers]
  • load
  • MTU

In EIGRP, metrics is represented by an integer from 0 to 4,294,967,295 (The size of a 32-bit integer). In Microsoft Windows XP routing it ranges from 1 to 9999.

A Metric can be considered as:[1]

  • additive - the total cost of a path is the sum of the costs of individual links along the path,
  • concave - the total cost of a path is the minimum of the costs of individual links along the path,
  • multiplicative - the total cost of a path is the product of the costs of individual links along the path.

A survey of routing metrics can be found here.

Service Level Metrics[edit]

Service Level Metrics are concerned with the end user's experience of using the product.

Availability[edit]

The availability of a computer network (or an individual service) may be expressed using the notation hh/d/ww. For a 24 hour service, seven days a week, available all year around, this would be expressed 24/7/52 (where the 52 stands for the number of weeks in a year). Service providers usually express that a service will be available for a percentage of this time.

To calculate the availability of a service expressed in this format, you need to do the following calculation:

98% availability on 24/7/52

  1. Multiply 24 hours per day by 7 days per week by 52 weeks per year = 8736 hours per year
  2. Find 98% of the hours per year = 8736 * 98 / 100 = 8561.28 hours guaranteed

You can then deduce how many full hours/days per year the service can be unavailable before the supplier is in breach of their Service Level Agreement. In this example, 8736 (hours) - 8561 (hours) = 175 hours (or around 7.3 days).

Notes[edit]

  1. ^ S. D. Rao, C. S. R. Murthy: “Distributed dynamic QoS-aware routing in WDM optical networks”, Computer Networks, Volume 48, Issue 4, 15 July 2005, Pages 585-604