Are there any mechanisms/techniques to implement bandwidth guarantees in packet switching networks?
Thank you!
2 Answers
While bandwidth guarantees can be implemented by policies within forwarding devices (routers, switches), most installations use a priority scheme instead (or in combination with bandwidth control). Bandwidth guarantees use reservations which are not as efficient as prioritizing by packet.
For Ethernet (layer 2), IEEE 802.1Q provides priority code points in an optional tag.
Likely the most commonly used are differentiated services for IP (layer 3) defined in RFC 2474.
Priorities in different layers can also be mapped to each other.
The options are:
- "Bandwidth limiting" heavy users by rate limiting
- QOS tagging
- packet dropping or delaying
Anything to reduce the traffic of others and give them less bandwidth to be used.
-
Just tagging with DSCP values for QoS does nothing. You need to configure the network devices to treat traffic based on the markings.– Ron Maupin ♦Mar 29, 2021 at 14:26
-
my idea was tagging packets with QOS labels on first hop and handle them on the next hop.– Max NMar 29, 2021 at 14:38
-
the man idea is to limit usage of other consumer to leave more bandwidth available for "important" traffic.– Max NMar 29, 2021 at 14:39
-
Simply tagging does nothing unless the network devices are configured to do something based on the tags (they are normally ignored), and there must be congestion on the network interface that is configured to treat the traffic based on the markings.– Ron Maupin ♦Mar 29, 2021 at 14:41
-
Would separate queues and rate limiting be the key techniques implemented on network devices to guarantee bandwidth, when tagging with DSCP values?– StarfishMar 30, 2021 at 3:22