One-way delay and instantanous packet
delay variation over time
CISCO 7200
In this test we verify the one-way delay and
ipdv evolution over time of a single EF stream
against different packet sizes and premium queue sizes (1 and 10 packs).
EF Stream profile:
Stream 1: 300 Kbps, variable packet size
Note: PVC bandwith is 1 Mbps.
Comments:
- Figure 3 shows that when no best-effort traffic
is injected the time properties of a EF stream complying with the EF
class are preserved. As with the IBM router one-way delay depends on
the packet size. However, unlike the corresponding
results on the IBM 2212,
for a given packet size one-way delay is almost constant, while with
the IBM one-way delay changed periodically.
- As expected, identical delay results are achieved when the PBS
increases from 1 to 10 packs as figure 5 shows.
- Instantaneous packet delay variation slightly increases with the
packet size but not in a significant way, as
figure 6 and figure 7.
show.
- When the PBS increases from 1 to 10 pack, only a few ipdv peaks
approximately increase from 250 usec to 270 usec. On the whole, we
can assume that the PBS is irrelevant, as the comparison of
figure 6 and figure 8 shows.
Figure 3: one-way delay evolution over time of an EF stream
over time for several packet sizes
(64, 128, 256, 1024, 1280 and 1518 bytes) and for PBS = 1 pack.
Figure 4: zoom on one-way delay evolution over time for packet size
equal to 1024 bytes.
Figure 5: same as figure 3 but with PBS = 10 pack.
Figure 6: ipdv over time for packet sizes of 64 and 256 bytes and
PBS = 1 pack.
Figure 7: ipdv over time for packet sizes of 1024 and 1518 bytes and
PBS = 1 pack.
Figure 8: same as figure 6 but with PBS = 10 packs.
Premium buffer
size does not impact the ipdv of the EF stream if departure_rate >
arrival_rate.
Last modified: Nov 24, 1999
