One-way delay and ipdv measurement methodology

One-way delay and instantaneous packet delay variation:
measurement methodology

The goal of this test is the estimation of two time metrics: one-way delay and istantanous packet delay variation for different values of a list of parameters list in the following.

One-way delay is intended according to RFC 2679:
for a real number dT, >>the *Type-P-One-way-Delay* from Src to Dst at T is dT<< means that Src sent the first bit of a Type-P packet to Dst at wire-time* T and that Dst received the last bit of that packet at wire-time T+dT.

Test parameters:

EF buffer size (PBS):
Premium buffer (IBM 2212) or queue-limit/tx-ring-limit (C7200)
queuing algorithm (WFQ or Priority Queueing -C7200-)
packet size: [64, 128, 256, 512, 1024, 1280, 1518] bytes (according to RFC 1242)
best-effort: with or without.
Best-effort profile: 200 pack/sec where each packet is of 1000 bytes (this corresponds to 2.0 Mbps of traffic, overhead included).

The SmartBits 200 was deployed as EF traffic source and sink.
The SmartMetrics test available from the SmartWindows v. 6.51 application was deployed for both one-way delay and ipdv measurement.
(Note: traffic has to be sinked to the same SmartBits in order to deploy the SmartWindows application.)

One-way delay is based on the cut-through latency measures collected by the SmartBits. This metric is measured according to RFC 1242, where cut-through latency is defined as:
the time interval starting when the end of the first bit of the input frame reaches the input port and ending when the start of the first bit of the output frame is seen on the output port.

So, one-way delay can be computed from cut-through latency by adding the transmission time of the packet, which is constant for a given packet size (see figure 2).

Figure 2: relationship between cut-through latency and one-way delay.

The timestamp deployed by the SmartBits represents the clock tick at the time of transmission (an absolute number) and the timestamp is added in hardware. The timestamp resolution is of 100 nsec.

The instantanous packet delay variation is computed by taking the abolute number of the ipdv metric defined by the IPPM working-group at IETF.

One-way delay over time is plotted by taking a given test and by plotting the one-way delay of each sample in packet sequence number order. Similarly, for ipdv over time the ipdv metric is computed and plotted according to the packet sequence number order.

Average one-way delay and average ippm for a given packet size are computed from the sample means over a set of repetitions of the same test.
Each test consists of approximately 250 samples.

References:

RFC 1242: Benchmarking Terminology for Network Interconnection Devices; S.Bradner
RFC 2679: A One-way Delay Metric for IPPM; G. Almes, S. Kalidindi, M. Zekauskas.

Last modified: Dic 12, 1999