Mehadi Hasan Mitul mmitul at banglalinkgsm.com
Mon Nov 18 09:51:24 BDT 2013

My opinion is same as of Roman vai.
Moving to the main topic of Simon vai, where the concentration is to fast convergence in large MPLS network. I think it can be achieved in the following ways: -
'Enable the MPLS applications for example: MPLS VPN, TE-FRR & MPLS Based QoS'.

a.     In MPLS VPN based network; OSPF will only manage the L3 uplinks between PE-P & P-P. Therefore the size of routing table will remain minimum. Use MP-iBGP to share the VPNv4 prefixes between PEs.

b.    Enable TE-FRR between P-P and PE-P; which will ensure the convergence in <50ms.

c.     Enable MPLS based QoS and assign traffic priority to required customer VPN only.

d.    Also vendor specific VPN-FRR technology can be used with the aid of BFD. This technology offers convergence between PE-PE in <200ms.

Mehadi Hasan Mitul
PS Core & MPBN Planning Team Leader
Cell: +880 19161 00368

From: nog-bounces at bdnog.org [mailto:nog-bounces at bdnog.org] On Behalf Of Nurul Islam
Sent: 18 November, 2013 8:35 AM
To: Md. Faridul Alam | Aamra; 'Simon Sohel Baroi / IIG-ITC / Sr.Manager / 01678618243 /'; nog at bdnog.org
Subject: Re: [bdNOG] ISIS of OSPF

With regards to IGP convergence issues I believe there is no big difference between OSPF and IS-IS if I am not wrong. My 1+ for IS-IS if you have multiple routed protocol I.e. IPv4 and IPv6 etc.

In SP network convergence issues are mostly handled in design level I.e. de-coupling IGP and EGP. I.e Only loopback and p2p link in OSPF/ISIS. So if there is any topology change in IGP, packet loss probability is very very less.

For external prefixes like downstream customer and upstream Internet, people mostly use eBGP. Across the infrastructure iBGP. Convergence time is an issue here specially if you have full BGP feed from multiple upstream. Try using vendor specific tools like BFD, BGP scanner etc. Or full BGP feed from single upstream and default from other etc. What else? :)



From: "Md. Faridul Alam | Aamra" <faridul.alam at aamra.com.bd<mailto:faridul.alam at aamra.com.bd>>
Date: Monday, 18 November 2013 5:49 AM
To: "'Simon Sohel Baroi / IIG-ITC / Sr.Manager / 01678618243 /'" <simon.baroi at fiberathome.net<mailto:simon.baroi at fiberathome.net>>, "nog at bdnog.org<mailto:nog at bdnog.org>" <nog at bdnog.org<mailto:nog at bdnog.org>>
Subject: Re: [bdNOG] ISIS of OSPF

I prefer IS-IS over OSPF for below reasons:

1.      Security

2.      Modularity

3.      Overload Mechanism
Also, I prefer IS-IS over OSPF because we can design a large networks by building a single large Level 1 (L1) area without any hierarchies in IS-IS and still be able to manage, something that would be difficult with OSPF. There are issues with inter-area traffic engineering and such and most people would like to keep their network as a single area if the routing protocol can manage it.

There are some restrictions for applications like MPLS TE that require flat area 0 for OSPF or flat level 2 for IS-IS. This also simplifies the design so heirarchy isn't an issue. But maintaining large number of LSA's in a single area is very difficult for OSPF. So IS-IS is better for MPLS.

Regards, Farid
Aamra, the power of WE
From: nog-bounces at bdnog.org<mailto:nog-bounces at bdnog.org> [mailto:nog-bounces at bdnog.org] On Behalf Of Simon Sohel Baroi / IIG-ITC / Sr.Manager / 01678618243 /
Sent: 18-Nov-2013 12:08 AM
To: nog at bdnog.org<mailto:nog at bdnog.org>
Subject: [bdNOG] ISIS of OSPF

Dare All the bdNOG friends,

Which one is better ISIS or OSPF.
I'm mainly focusing of fast convergence time. My clients can not give me any ping loss in that convergence moment, if one fiber goes down in my MPLS link.
Number of router will be around 600.
Need your help guys.
Fiber at Home.

Simon Sohel Baroi  |  Sr. Manager, Technology  |  PICO  |   ITC - IIG  |
Cell : +880-1678-618243, +880-181-7022207  |  Desk : +880-9666776677 Ext-1031  |
Mail : simon.baroi at pico.net.bd<mailto:simon.baroi at fiberathome.net>  |  Skype : tx.fttx  |
Reduce. Reuse. Recycle. Respect. It's the little things that really can make a difference.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.bdnog.org/pipermail/nog/attachments/20131118/56324f65/attachment-0001.html>

More information about the nog mailing list