Avaya Configuring PPP Services Manual de usuario Pagina 40

  • Descarga
  • Añadir a mis manuales
  • Imprimir
  • Pagina
    / 118
  • Tabla de contenidos
  • MARCADORES
  • Valorado. / 5. Basado en revisión del cliente
Vista de pagina 39
Configuring PPP Services
2-14 114068 Rev. A
Balancing Traffic Loads
In a configuration with multilink enabled, a sending router divides the outbound
traffic among all the lines in the bundle. The configured external clock speed of
each line determines the proportion of the total traffic each receives. For example,
pairing a 9600-bit/second line with an ISDN-B channel yields a clock-speed ratio
of roughly 1:6.8, assuming same-size packets. That is, for every packet sent on the
slower link, the router can send about seven packets on the faster link.
On the receiving end, multilink resequences packets arriving on different links
using the sequence number from the multilink header. Gaps in the ordering may
occur, however, when packets are corrupted or otherwise lost or when they arrive
after packets with later sequence numbers. To minimize this situation, multilink
buffers out-of-sequence packets in case the preceding sequence-numbered packets
arrive shortly after the later-numbered packets.
Using Multilink Fragmentation
By default, PPP multilink allows packet fragmentation. With fragmentation
enabled, PPP splits large datagrams into smaller packets and sends these packets
across links in a multilink bundle. Enabling fragmentation means that PPP can
split packets when necessary for better performance. PPP does not arbitrarily split
all the packets it transmits. Fragmentation improves the distribution of data across
multilink lines and uses buffer resources more efficiently, thereby improving the
flow of data over multilink circuits. By default, multilink fragmentation is
enabled. Bay Networks routers comply with RFC 1717, which defines PPP
multilink.
Without multilink fragmentation, when PPP sends packets over a multilink
bundle, it sends one packet over each line in sequence in a round-robin fashion. To
optimize performance, PPP does attempt to send fewer packets on the slower
lines. However, some packets with higher sequence numbers sent over faster lines
could be received earlier than packets with lower sequence numbers sent over
slower lines. Since PPP maintains the sequence of received packets, the receiving
peer must store the out-of-sequence packets until the delayed packet arrives, and
this can result in slower network performance. If the number of packets needing to
be resequenced is greater than the available allowed buffer space, some packets
could be considered late and discarded.
Vista de pagina 39
1 2 ... 35 36 37 38 39 40 41 42 43 44 45 ... 117 118

Comentarios a estos manuales

Sin comentarios