Re: [c-nsp] Anycast GW for L2 subnet
Hi Tim, > Adam Vitkovsky IP Engineer T: 0333 006 5936 E: adam.vitkov...@gamma.co.uk W: www.gamma.co.uk This is an email from Gamma Telecom Ltd, trading as “Gamma”. The contents of this email are confidential to the ordinary user of the email address to which it was addressed. This email is not intended to create any legal relationship. No one else may place any reliance upon it, or copy or forward all or any of it in any form (unless otherwise notified). If you receive this email in error, please accept our apologies, we would be obliged if you would telephone our postmaster on +44 (0) 808 178 9652 or email postmas...@gamma.co.uk Gamma Telecom Limited, a company incorporated in England and Wales, with limited liability, with registered number 04340834, and whose registered office is at 5 Fleet Place London EC4M 7RD and whose principal place of business is at Kings House, Kings Road West, Newbury, Berkshire, RG14 5BY. -Original Message- > From: Tim Stevenson [mailto:tstev...@cisco.com] > Sent: 07 August 2015 16:28 > A couple caveats: > - in theory you could have more sites, each with an AC pair, but we > only support 4 AC gwys today That is interesting as all the docs suggests that there's no limit "n-way active gateway" Where does the limitation come from please? I guess there's no limit on the number of switches that can advertise the same ASID and also I'm not aware of any upper limit of maximum members per HSRP group. > - FP still builds the MD trees as it always has, ie you avoid > tromboning of unicast, but multidestination (bc/mc/uuc) still has the > potential to 'take the scenic route' through the other site, > depending on the topology of each tree > Interesting point I'll definitely keep an eye on this one. Though I expect the BUM traffic to be flooded everywhere since the VLANs span across both locations. adam ___ cisco-nsp mailing list cisco-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
Re: [c-nsp] Anycast GW for L2 subnet
At 01:25 AM 8/7/2015 Friday, Adam Vitkovsky quipped: <...snip...> > You could do that but the idea with anycast HSRP > is that all participating HSRP routers equally > distribute the L3 switching load. > Yes I agree but in my case the DC would span across two geographic locations and so would the VLANs. So I was thinking I could enforce VMs in either location to primarily use local GWs as the closest exit from the VLAN. So kind of L2 routing or anycasting of the GWs. When I do the same in L3 i.e. anycast the VLAN prefix I should get a nice split and no tromboning of traffic between the locations. <...snip...> Yes, this will work from an Anycast HSRP standpoint, ie, if you have two sites, two AC routers at each, leaf switches in each site would only vector gwy traffic to the 2 local (closest) AC routers. A couple caveats: - in theory you could have more sites, each with an AC pair, but we only support 4 AC gwys today - FP still builds the MD trees as it always has, ie you avoid tromboning of unicast, but multidestination (bc/mc/uuc) still has the potential to 'take the scenic route' through the other site, depending on the topology of each tree We are actually working on solutions to both those limitations but I am speaking only of what we have today. Hope that helps, Tim Tim Stevenson, tstev...@cisco.com Routing & Switching CCIE #5561 Distinguished Engineer, Technical Marketing Data Center Switching Cisco - http://www.cisco.com +1(408)526-6759 ___ cisco-nsp mailing list cisco-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
Re: [c-nsp] Anycast GW for L2 subnet
Hi Tim, Thank you very much for your explanation, I really appreciate it. > Tim Stevenson > Sent: 05 August 2015 02:57 > > Hi Adam, please see inline below: > > At 03:58 PM 8/4/2015 Tuesday, Adam Vitkovsky quipped: > >Has anyone played with Anycast HSRP with fabric path please? > >Just would like to confirm I understand it correctly. > >So ISIS calculates best path to the anycast switch ID advertising the HSRP > MAC > > > There are no MAC advertisements in FP, routing is based on switch IDs (SID). Aah yes, sure, must have thinking evpn :) Right the MAC learning is done in data-plane. > > and since I can manipulate metrics on links > > between spine and leaf switches I should be > > able to dictate which leaf switches should be using which GWs right? > > > You could do that but the idea with anycast HSRP > is that all participating HSRP routers equally > distribute the L3 switching load. > Yes I agree but in my case the DC would span across two geographic locations and so would the VLANs. So I was thinking I could enforce VMs in either location to primarily use local GWs as the closest exit from the VLAN. So kind of L2 routing or anycasting of the GWs. When I do the same in L3 i.e. anycast the VLAN prefix I should get a nice split and no tromboning of traffic between the locations. > >Because only paths to anycast switch ID with > >equal costs are considered for multipathing > >right? (i.e. there’s no unequal cost load sharing correct?) > > Correct, it is ECMP only. > > The model is that all anycast HSRP routers have > their own unique SID but also an emulated SID > shared among them all. All advertise that ESID, > and any FP switch with equal path cost to 2 or > more of those will load balance traffic destined to the HSRP MAC among > them. > > Typical topology is spine/leaf but any topology > will work. Note that only the control-plane > Active router is the one that responds to ARP & > sources HSRP hellos with the HSRP MAC (using the > ESID as the source SID in FP frames). > I see now, that's how everyone associates the HSRP MAC with the ESID and can start loadsharing. Thank you very much. > See section 10 here for a bit more: > http://www.cisco.com/c/en/us/products/collateral/switches/nexus-7000- > series-switches/white_paper_c11-687554.html > > > Hope that helps, > Tim > > > > > > Tim Stevenson, tstev...@cisco.com > Routing & Switching CCIE #5561 > Distinguished Engineer, Technical Marketing > Data Center Switching > Cisco - http://www.cisco.com > +1(408)526-6759 > > > Adam Vitkovsky IP Engineer T: 0333 006 5936 E: adam.vitkov...@gamma.co.uk W: www.gamma.co.uk This is an email from Gamma Telecom Ltd, trading as “Gamma”. The contents of this email are confidential to the ordinary user of the email address to which it was addressed. This email is not intended to create any legal relationship. No one else may place any reliance upon it, or copy or forward all or any of it in any form (unless otherwise notified). If you receive this email in error, please accept our apologies, we would be obliged if you would telephone our postmaster on +44 (0) 808 178 9652 or email postmas...@gamma.co.uk Gamma Telecom Limited, a company incorporated in England and Wales, with limited liability, with registered number 04340834, and whose registered office is at 5 Fleet Place London EC4M 7RD and whose principal place of business is at Kings House, Kings Road West, Newbury, Berkshire, RG14 5BY. ___ > cisco-nsp mailing list cisco-nsp@puck.nether.net > https://puck.nether.net/mailman/listinfo/cisco-nsp > archive at http://puck.nether.net/pipermail/cisco-nsp/ adam ___ cisco-nsp mailing list cisco-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
Re: [c-nsp] Anycast GW for L2 subnet
Hi Adam, please see inline below: At 03:58 PM 8/4/2015 Tuesday, Adam Vitkovsky quipped: Has anyone played with Anycast HSRP with fabric path please? Just would like to confirm I understand it correctly. So ISIS calculates best path to the anycast switch ID advertising the HSRP MAC There are no MAC advertisements in FP, routing is based on switch IDs (SID). and since I can manipulate metrics on links between spine and leaf switches I should be able to dictate which leaf switches should be using which GWs right? You could do that but the idea with anycast HSRP is that all participating HSRP routers equally distribute the L3 switching load. Because only paths to anycast switch ID with equal costs are considered for multipathing right? (i.e. thereâs no unequal cost load sharing correct?) Correct, it is ECMP only. The model is that all anycast HSRP routers have their own unique SID but also an emulated SID shared among them all. All advertise that ESID, and any FP switch with equal path cost to 2 or more of those will load balance traffic destined to the HSRP MAC among them. Typical topology is spine/leaf but any topology will work. Note that only the control-plane Active router is the one that responds to ARP & sources HSRP hellos with the HSRP MAC (using the ESID as the source SID in FP frames). See section 10 here for a bit more: http://www.cisco.com/c/en/us/products/collateral/switches/nexus-7000-series-switches/white_paper_c11-687554.html Hope that helps, Tim Tim Stevenson, tstev...@cisco.com Routing & Switching CCIE #5561 Distinguished Engineer, Technical Marketing Data Center Switching Cisco - http://www.cisco.com +1(408)526-6759 ___ cisco-nsp mailing list cisco-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
[c-nsp] Anycast GW for L2 subnet
Hi folks, It is interesting that there seem to be only two options to accomplish anycast GWs for the L2 subnet with selection based on closest proximity to a GW (would have thought there’re plenty of solutions out there with all the DC hype dating several years back). I only found Anycast HSRP with fabric path and the Distributed Anycast GW for VXLAN with MP-BGP EVPN Control Plane –but that one seem to be only for the big boys :) Has anyone played with Anycast HSRP with fabric path please? Just would like to confirm I understand it correctly. So ISIS calculates best path to the anycast switch ID advertising the HSRP MAC and since I can manipulate metrics on links between spine and leaf switches I should be able to dictate which leaf switches should be using which GWs right? Because only paths to anycast switch ID with equal costs are considered for multipathing right? (i.e. there’s no unequal cost load sharing correct?) Thank you adam Adam Vitkovsky IP Engineer T: 0333 006 5936 E: adam.vitkov...@gamma.co.uk W: www.gamma.co.uk This is an email from Gamma Telecom Ltd, trading as “Gamma”. The contents of this email are confidential to the ordinary user of the email address to which it was addressed. This email is not intended to create any legal relationship. No one else may place any reliance upon it, or copy or forward all or any of it in any form (unless otherwise notified). If you receive this email in error, please accept our apologies, we would be obliged if you would telephone our postmaster on +44 (0) 808 178 9652 or email postmas...@gamma.co.uk Gamma Telecom Limited, a company incorporated in England and Wales, with limited liability, with registered number 04340834, and whose registered office is at 5 Fleet Place London EC4M 7RD and whose principal place of business is at Kings House, Kings Road West, Newbury, Berkshire, RG14 5BY. ___ cisco-nsp mailing list cisco-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/