IRC Logs

Network: freenode Channel: #cisco

	Enter your search terms Submit search form
	Web www.irclog.org

[02:32]<znzwzzn>well routes
[02:32]<cuvyncjmnz>jerware: no, I mean fiber / copper carrier. If the fiber/copper gets cut, well, only 1 circuit gets fucked up the goat ass
[02:33]<xud2jfv>CutieCoder: tell me more. Your concern is redundency, right.
[02:33]<znzwzzn>right, redundancy
[02:34]<cuvyncjmnz>jerware: resilliency (different than redundancy), redundancy, symmetry and proper routing is all I care for
[02:35]<znzwzzn>CutieCoder: well if they are both intact will traffic be spill up equally between both T1 interfaces?
[02:35]<znzwzzn>sounds like you have done this before with T1
[02:35]<cuvyncjmnz>jerware: no, that depends on your routing and load balancing
[02:35]<znzwzzn>oh ok
[02:36]<znzwzzn>like setting the cost, and bandwith with the proper commands and such
[02:36]<znzwzzn>delay, etc etc
[02:36]<cuvyncjmnz>jerware: in addition, you must not have any exit of last resort for that to be done properly. BGP / MPLS baby ;)
[02:36]<fjzvxnz>you could do ip load-sharing per-packet or similar, but i'm not sure how well that would work on interfaces from different ISPs.
[02:36]<cuvyncjmnz>jerware: I mean, lets say you have multiple circuit types
[02:36]<znzwzzn>i dont know BGP or MPLS, im ccna level
[02:36]<cuvyncjmnz>Ohhh god
[02:36]<cuvyncjmnz>Nevermind ;)
[02:37]<znzwzzn>:P sorry
[02:37]<xud2jfv>one more question to my multilink: don't multilinks have significantly more latency anyway?
[02:37]<cuvyncjmnz>jerware: it's fine
[02:37]<cuvyncjmnz>jerware: I'll explain
[02:38]<cuvyncjmnz>jerware: see, take for example that you have 2 circuits. 1 fastE and one gigE.
[02:38]<vcul>what does it mean when a bri interface shows as "spoofing"?
[02:38]<znzwzzn>ok
[02:38]<cuvyncjmnz>jerware: the fastE circuit is unmetered (fixed commit pipe). the gigE circuit is capped at lets say, 300mbit and you have 10TB of quota.
[02:39]<fjzvxnz>humbolt: because of the encapsulation overhead, generally yes.
[02:40]<cuvyncjmnz>jerware: you would want to build your routing to ensure that traffic coming in and out for lets say pr0n servers to use the fastE circuit since you won't pay more out of it
[02:41]<cuvyncjmnz>jerware: for the "web hosting" clients (shitty website plans, you know, stuff that doesn't drink bandwidth for shit), your routing should be set to come in/out the gigE circuit
[02:42]<znzwzzn>but isnt gigE faster than fastE ?
[02:42]<cuvyncjmnz>gigE can carry more, but we are dealing with restrictions here.
[02:43]<cuvyncjmnz>you got a quota to deal with in this scenario
[02:43]<cuvyncjmnz>Get it?
[02:43]<`kgvyg>jerware: Cutie is right, if you start to get your fastE to 75% or so, then move some traffic to the gigE
[02:44]<cuvyncjmnz>Exactly
[02:44]<cuvyncjmnz>change your route-map to influence the routing for prefix XYZ/21 to fly through the other exit disc
[02:46]<cuvyncjmnz>GNIIII
[02:48]<cuvyncjmnz>some idiot is gonna die soon at work... fucking with my label stacks is not recommended.
[02:49]<gndyvx>fight them
[02:49]<cuvyncjmnz>600V tri-phase killing
[02:49]<gndyvx>as an alternative to route-maps, may i recommend VRFs
[02:49]<gndyvx>clean up the code alot and more secure
[02:49]<cuvyncjmnz>nemith++
[02:50]<gndyvx>whats up CutieCoder
[02:50]<gndyvx>CutieCoder: are you the same sarah that was in here before but with a new nick?
[02:50]<cuvyncjmnz>nemith: most of them won't use MPLS/CEF/VRF/VPNv4 :P
[02:50]<cuvyncjmnz>nemith: no I'm not
[02:50]<gndyvx>oh well then, welcome to the chan
[02:51]<cuvyncjmnz>nemith: same ol' going on... in vacation, being disturbed because some asshats down at 60 hudson can't admin for shit
[02:51]<cuvyncjmnz>nemith: thank you dear ;)
[02:51]<gndyvx>yeah.. i've done little mpls. but I have found a lot of solutions using vrf-light
[02:51]<cuvyncjmnz>let alone (g)MPLS
[02:51]<gndyvx>yeah
[02:52]<znzwzzn>nemith: what did Sarah do?
[02:52]<gndyvx>jerware: nothing
[02:52]<znzwzzn>that she had to come back in a different nick
[02:52]<znzwzzn>oh ok
[02:52]<gndyvx>I haven't seen her in a while, just curious
[02:52]<znzwzzn>ahh
[02:53]<cuvyncjmnz>A full blown MPLS architecture, if not broken by design -- makes your life, scallability and ressource allocation so easy
[02:53]<gndyvx>yeah.. I am makeing a buisness case to implement a VRF cloud
[02:53]<gndyvx>er MPLS cloud
[02:54]<gndyvx>it even makes sense for enterprises
[02:54]<cuvyncjmnz>nemith: teh VRF gets to your head? :)
[02:54]<cuvyncjmnz>hehe
[02:54]<gndyvx>heh
[02:54]<gndyvx>like mug rootbeer
[02:54]<cuvyncjmnz>good idea
[02:54]<cuvyncjmnz>BARQS!
[02:54]<gndyvx>heh
[02:54]<cuvyncjmnz>has bite
[02:55]<fjzvxnz>hmm, i could go for s/root// right now :D
[02:55]<gndyvx>heh
[02:55]<fjzvxnz>it's almost 5.
[02:55]<gndyvx>yeah i will be doing that later on
[02:55]<fjzvxnz>i'm *finally* off-call this weekend
[02:55]<fjzvxnz>that 24/7/365 shit was starting to get to me.
[02:56]<gndyvx>heh
[02:56]<cuvyncjmnz>beeer
[02:56]<znzwzzn>mmm
[02:56]<afsc>When a router switch a packet from a network to another network does it strip the ethernet header or does it pass it to the next network too?
[02:56]<afsc>and checksum
[02:57]<gndyvx>the only thing modified is the TTL
[02:57]<mrxmjum-rpc>hmmm
[02:57]<afsc>so ethernet headers are sent across the internet?
[02:57]<znzwzzn>so the same ethernet header is maintated between routers with ethernet
[02:57]<mrxmjum>netacad says, switchport port-security mac-address sticky can't be used in voice vlans.. any idea why?
[02:58]<sag>jerware: no.
[02:58]<znzwzzn>oh
[02:58]<sag>or. between?
[02:58]<gndyvx>ethernet headers are not
[02:58]<cuvyncjmnz>brb
[02:58]<gndyvx>ip headers are
[02:59]<mrxmjum>nemith, aren't they dot1q or isl encapsulated?
[02:59]<znzwzzn>between hops [rtr]----eth-----[rtr]------eth-----[rtr]-----eth------[rtr]. so for each hop, a new ethernet header is stripped, and a new one is encapsulated. correct?
[02:59]<mrxmjum>jerware, yea
[02:59]<znzwzzn>ahh
[02:59]<gndyvx>yes
[02:59]<mrxmjum>jerware, but not exactly a new one, they may use some fields from previous ones
[03:00]<gndyvx>it could be another layer medium as well

Back to #cisco
Or go to some related logs:
#blender
#gentoo
#ubuntu
#gentoo
#azureus