[j-nsp] JUNIPER POLICER and CoS Shaping Rate

OBrien, Will ObrienH at missouri.edu
Wed Oct 3 23:35:01 EDT 2012


The math for burst rate is a little odd. However it seems to average at around 10% of the desired rate. The burstiness (for lack of a better word) provides for a better user experience rather than a hard policer.

Will O'Brien

On Oct 3, 2012, at 9:55 PM, "GIULIANO (WZTECH)" <giuliano at wztech.com.br> wrote:

> People,
> 
> Some topics where questioned today about how to limit traffic for vlan subscribers using MX5 routers.
> 
> The main question is related to system architecture related to the main gear (internal machine) to control and limiting packets.
> 
> Using policers (input or output) or shaping-rate we have quite the same result: miscalculating or error.
> 
> If we create a rule like the following:
> 
> 
> set class-of-service interfaces ge-0/0/1 unit 530 shaping-rate 20m
> 
> 
> The output traffic rates 19.2~ Mbps only (using MRTG and SNMP statistics and graphics).
> 
> We ever needs to allocate more bandwidth for the subscriber like.
> 
> set class-of-service interfaces ge-0/0/1 unit 530 shaping-rate 22m
> 
> To get the correct result ...
> 
> Using policers generate almost the same result for output traffic.
> 
> Is this because of system architecture or this is a graphic's mistake ?
> 
> The burst size limit influence this result ? It must be calculated using what kind of parameter ?
> 
> For example (same physical interface, same MTU, etc):
> 
> Interface ge-0/0/0 unit 10 - VLAN 10 - 30 Mbps What is the correct burst ?
> 
> Interface ge-0/0/0 unit 20 - VLAN 20 - 50 Mbps What is the correct burst ?
> 
> Interface ge-0/0/0 unit 30 - VLAN 30 - 150 Mbps What is the correct burst ?
> 
> Interface ge-0/0/0 unit 30 - VLAN 30 - 4 Mbps What is the correct burst ?
> 
> Does anyone has solved this problems ?
> 
> Is it possible to get a correct parameter and points to a correct limit for the contracted bandwidth ?
> 
> Thanks a lot,
> 
> Giuliano
> 
> 
> _______________________________________________
> juniper-nsp mailing list juniper-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/juniper-nsp



More information about the juniper-nsp mailing list