Re: [j-nsp] PIC Oversubscription Show Commands
Yes if traffic goes in/out of the same PIC. We have some IQ2Es on some M10is. The numbers were 4:1 when testing with a Spirent, but smaller MTU size packets @ 64 and 128 the transfer rate is less. Can't recall any commands for the PIC, just recall seeing no drops on the interface queues and knowing the loss was due to the traffic flow oversubscribing the PIC. Thanks, Eduardo Eduardo Barrios, EIT, JNCIP-SP Telecommunications Specialist Lower Colorado River Authority | 3505 Montopolis Dr. | Austin, TX 78744 512.730.6332 ph -Original Message- From: juniper-nsp [mailto:juniper-nsp-boun...@puck.nether.net] On Behalf Of mdb1986 Sent: Wednesday, February 19, 2014 4:30 PM To: juniper-nsp@puck.nether.net Subject: [External] [j-nsp] PIC Oversubscription Show Commands My understanding is the 4x1GE IQ2 PIC is oversubscribed 4:1. Are there any commands to see if there is congestion on the PIC? Really, I'm wondering if there are any commands that show congestion within the switch itself (ie PIC to/from FPC, FPC to/from CFEB, etc.,) I can't seem to find any. Thanks, ___ juniper-nsp mailing list juniper-nsp@puck.nether.netmailto:juniper-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/juniper-nsp ___ juniper-nsp mailing list juniper-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/juniper-nsp
[j-nsp] ECMP in Virtual Router and IP-Track
Greetings Mytel, Regarding your IP tracking question, the feature is called IP Monitoring with Route Fail-over: http://kb.juniper.net/InfoCenter/index?page=contentid=KB25052 http://pathfinder.juniper.net/feature-explorer/feature-info.html?fKey=4037 AFAICT it's only supported on branch SRX at the moment, but you could certainly try it on your bigger boxes with a recent 12.1X build. If it's not there then you really should bug your account team about it as this is one of the but Crisco does it! features missing from JunOS for a long time (along with CLI aliases that don't eat CPU by executing a script...). Of course they also suggest using this scripted mess if your platform doesn't support the new feature: http://www.juniper.net/us/en/community/junos/script-automation/library/event/track-ip/ Probably works fine for a few routes but I wouldn't want to make a habit of using that or roll it out at scale... Best of luck, --- Blake Willis Network Engineering Consultant Scalable System Design LLC ___ juniper-nsp mailing list juniper-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/juniper-nsp
Re: [j-nsp] VLAN's on EX4300 with 13.2X50-D15.3
On 20 Feb 2014, at 6:00 pm, Per Granath per.gran...@gcc.com.cy wrote: For the mixed VC, the options are EX42+EX4550 or EX43+QFX. For VC, the EX42 uses VCP-cables, while the EX43 uses QSFP-DAC-cables (assuming you do not want to waste your 10G ports). EX42 and EX43 has the same price. EX4550 is somewhat cheaper than QFX. A word of caution on the same price between EX42/43: If you want to use the EX4300 for L3 (and lets face it, that's probably everyone) you'll require the Enhanced Feature License for OSPFv2/3, IGMP v1/v2, PIM, VRF-Lite, QinQ and OAM. EX4200 includes all this in the base license. Both still require the Advanced Feature License for BGP, ISIS and MPLS. If you want to install the Advanced Feature License on EX4300 though, you need to buy both the EFL *and* AFL. All features are still available for testing without licenses, but you get the nag message every minute in syslog. All that said, being able to VC up to 5 of these things together with a dedicated 40G between any two is pretty darn cool. Ben ___ juniper-nsp mailing list juniper-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/juniper-nsp
Re: [j-nsp] VLAN's on EX4300 with 13.2X50-D15.3
weren't the ex4200 VC connections 64/128 Gbps thru the ribbon cable? why is 40G which uses up actual ports all that exciting? maybe i don't see it because it doesn't apply to my architecture. :-/ ___ juniper-nsp mailing list juniper-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/juniper-nsp
Re: [j-nsp] VLAN's on EX4300 with 13.2X50-D15.3
The VC cables in EX4200s are 32G half duplex. If we go full duplex we get to 64. Add another VC cable and we get 128. With the 40G interface, we get 80 full duplex and 160 with 2. HTH On 20 February 2014 22:31, ryanL ryan.lan...@gmail.com wrote: weren't the ex4200 VC connections 64/128 Gbps thru the ribbon cable? why is 40G which uses up actual ports all that exciting? maybe i don't see it because it doesn't apply to my architecture. :-/ ___ juniper-nsp mailing list juniper-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/juniper-nsp ___ juniper-nsp mailing list juniper-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/juniper-nsp
[j-nsp] packets reclassification basing on src/dst ASN
Hello! I'm trying to do some packet reclassification basing on source / destination ASN. I've tried sample I've found on stackexchange on my lab MX960/12.3R3-S1 set class-of-service forwarding-policy class POLICY-1 classification-override forwarding-class network-control set policy-options policy-statement MARK-POLICY-1 term term-1 to as-path BGP-PATH-TO-MARK-COS set policy-options policy-statement MARK-POLICY-1 term term-1 then class POLICY-1 set routing-options forwarding-table export MARK-POLICY-1 set policy-options as-path BGP-PATH-TO-MARK-COS 65512 .* set class-of-service interfaces ge-5/2/0 unit 0 rewrite-rules inet-precedence default There are two ASN peers with that router - 65512, 65513, 65510 is router itself. Example works as expected - packet to AS65512 get their tos rewritten and tcpdump shows that. However, if I try to revers a logic and want to remark all packets to 65512 that's does not work: set policy-options policy-statement MARK-POLICY-1 term term-1 from as-path BGP-PATH-TO-MARK-COS Looks like nothing but '.*' matches for 'from' stanza. Tried the same with the prefixes - same story: 'to' - worked , 'from' - not. What could be wrong ? ___ juniper-nsp mailing list juniper-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/juniper-nsp
Re: [j-nsp] VLAN's on EX4300 with 13.2X50-D15.3
On Thu, Feb 20, 2014 at 02:31:58PM -0800, ryanL wrote: weren't the ex4200 VC connections 64/128 Gbps thru the ribbon cable? why is No, 32 Gbps through PCIe x8. Marketing would have you believe it is 128 Gbps, but using their definitions, a 10 GigE port is 40 Gbps. In reality the VCPs each do 32 Gbps full-duplex, and there are two of them per switch. So 2*(32 in + 32 out) = 128 Gbps. 40G which uses up actual ports all that exciting? maybe i don't see it because it doesn't apply to my architecture. :-/ 40 Gbps is bigger than 32 Gbps? Standard cabling? ___ juniper-nsp mailing list juniper-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/juniper-nsp