Improvements for the TBF chapter
PATCH 1/1 Improvements for the TBF chapter
Hi
Please consider applying this patch:
- Clearifies the relation between 'rate' and 'burst'
- Corrects statement about non-existing input qdisc (-> IMQ)
bye,
-christian-
Index: lartc.db
=================================RCS file: /var/cvsroot/2.4routing/lartc.db,v retrieving revision 1.12
lartc.db +21 -17
@@ -2102,18 +2102,27 @@ </Para> </VarListEntry> <VarListEntry> -<Term>burst/buffer/maxburst</Term> +<Term>rate and burst/buffer/maxburst</Term> <ListItem> <Para> -Size of the bucket, in bytes. This is the maximum amount of bytes that -tokens can be available for instantaneously. In general, larger shaping -rates require a larger buffer. For 10mbit/s on Intel, you need at least -10kbyte buffer if you want to reach your configured rate! +Rate is the amount of tokens that are tried to put into the bucket every +second specified in mbit/s. +</Para> +<Para> +Burst is the size of the bucket, in bytes. This is the maximum amount of bytes +that tokens can be available for instantaneously. </Para> - <Para> -If your buffer is too small, packets may be dropped because more tokens -arrive per timer tick than fit in your bucket. +In general, larger shaping rates require a larger buffer. Remember that if +your buffer is too small, packets may be dropped because regardless how +much tokens the 'rate' parameter tries to put in the bucket, the ones that +actually fit into it are not sufficient for all waiting data. +</Para> +<Para>To archive a throughput of 10mbit/s on Intel you have to set the buffer +to 12.8kbyte. Calculation: On Intel the kernel HZ variable is 100, which means +that the token filler function is called 100 times per second. 100 times +12.8kbyte are 1310720byte or 10485760bit that are pushed into the bucket - +assuming a high enough value for the rate variable. </Para> </VarListEntry> <VarListEntry>
@@ -2125,13 +2134,6 @@ token usage for a packet. </Para> </VarListEntry> -<VarListEntry> -<Term>rate</Term> -<ListItem> -<Para> -The speedknob. See remarks above about limits! -</Para> -</VarListEntry> </VariableList> </Para>
@@ -2502,8 +2504,10 @@ <ListItem> <Para> Delaying or dropping packets in order to make traffic stay below a -configured bandwidth. In Linux, policing can only drop a packet and not -delay it - there is no 'ingress queue'. +configured bandwidth. In Linux, policing used to be only able to drop a packet +and not delay it - there was no 'ingress queue'. A currently developed solution +is the 'intermediate queueing device' (IMQ). It is discussed in a seperate +chapter. </Para> </VarListEntry> <VarListEntry>
Signature · 7 lines
Christian Hammers WESTEND GmbH - Aachen und Dueren Tel 0241/701333-0 [email protected] Internet & Security for Professionals Fax 0241/911879 WESTEND ist CISCO Systems Partner - Authorized Reseller _______________________________________________ LARTC mailing list / [email protected] http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/