Date: Fri, 28 May 1999 20:02:57 +0900 From: Yasuhiro Ohara <yasu@sfc.wide.ad.jp> To: ANDREAS.KLEMM.AK@bayer-ag.de Cc: andreas@klemm.gtn.com, freebsd-net@freebsd.org, zebra@zebra.org, kunihiro@zebra.org Subject: Re: Antwort: Re: [zebra 553] OSPF eequal-cost paths, which algor Message-ID: <19990528200257W.yasu@sfc.wide.ad.jp> In-Reply-To: Your message of "Fri, 28 May 1999 12:46:56 %2B0200" <0006800011799121000002L012*@MHS> References: <0006800011799121000002L012*@MHS>
next in thread | previous in thread | raw e-mail | index | archive | help
I'm sorry I don't know, but Mr. Alex Bligh have answered this... yasu From: ANDREAS.KLEMM.AK@bayer-ag.de Subject: Antwort: Re: [zebra 553] OSPF eequal-cost paths, which algor Date: Fri, 28 May 1999 12:46:56 +0200 ANDREAS.KLEMM.AK> ANDREAS.KLEMM.AK> Hi ! ANDREAS.KLEMM.AK> ANDREAS.KLEMM.AK> [sorry for junk characters, only have Lotus notes here, no better mailer] ANDREAS.KLEMM.AK> ANDREAS.KLEMM.AK> Thanks for answering. O.K. this would explain why the OSPF V2 RFC only ANDREAS.KLEMM.AK> explains, that there will be ANDREAS.KLEMM.AK> one routing entries with 2 paths in the OSPF routing table ... ANDREAS.KLEMM.AK> ANDREAS.KLEMM.AK> Do you perhaps know, how CISCO*%s use that information ? ANDREAS.KLEMM.AK> ANDREAS.KLEMM.AK> Do I get 4 or 8 MBit throughput in one direction ? ANDREAS.KLEMM.AK> ANDREAS.KLEMM.AK> Andreas /// ANDREAS.KLEMM.AK> ANDREAS.KLEMM.AK> ANDREAS.KLEMM.AK> ANDREAS.KLEMM.AK> ANDREAS.KLEMM.AK> ANDREAS.KLEMM.AK> yasu@sfc.wide.ad.jp on 28.05.99 10:54:54 ANDREAS.KLEMM.AK> ANDREAS.KLEMM.AK> An: kunihiro@zebra.org @ INTERNET, zebra@zebra.org @ INTERNET, ANDREAS.KLEMM.AK> freebsd-net@freebsd.org @ INTERNET, Andreas Klemm@BAYERNOTES, ANDREAS.KLEMM.AK> andreas@klemm.gtn.com @ INTERNET ANDREAS.KLEMM.AK> Kopie: ANDREAS.KLEMM.AK> Thema: Re: [zebra 553] OSPF eequal-cost paths, which algorithm, how ANDREAS.KLEMM.AK> ANDREAS.KLEMM.AK> ANDREAS.KLEMM.AK> I think... ANDREAS.KLEMM.AK> ANDREAS.KLEMM.AK> ANDREAS.KLEMM.AK> From: Andreas Klemm <andreas@klemm.gtn.com> ANDREAS.KLEMM.AK> Subject: [zebra 553] OSPF eequal-cost paths, which algorithm, how exactly load ANDREAS.KLEMM.AK> balancing ? Date: Fri, 28 May 1999 08:40:02 +0200 ANDREAS.KLEMM.AK> ANDREAS.KLEMM.AK> andreas> Hi ! ANDREAS.KLEMM.AK> andreas> ANDREAS.KLEMM.AK> andreas> I'm looking for the algorithm how OSPF does a routing decision and ANDREAS.KLEMM.AK> andreas> what kind of load balancing is been done, between the two 4 MBit ANDREAS.KLEMM.AK> andreas> leased lines, when OSPF has equal-cost paths like this. ANDREAS.KLEMM.AK> andreas> Cisco IOS 11.2 or 11.3. ANDREAS.KLEMM.AK> andreas> ANDREAS.KLEMM.AK> andreas> **Rstart ANDREAS.KLEMM.AK> andreas> **|e0 ANDREAS.KLEMM.AK> andreas> **|10MBit ANDREAS.KLEMM.AK> andreas> **| ANDREAS.KLEMM.AK> andreas> -----------+---------------- ANDREAS.KLEMM.AK> andreas> *|10MBit**|10MBit ANDREAS.KLEMM.AK> andreas> *|e0**|e0 ANDREAS.KLEMM.AK> andreas> |**| ANDREAS.KLEMM.AK> andreas> *R1**R3 ANDREAS.KLEMM.AK> andreas> ||**|| ANDREAS.KLEMM.AK> andreas> s0||s1 s0||s1**load balancing between 2 serial ANDREAS.KLEMM.AK> andreas> *||2x2=4MBit*||2x2=4MBit*(4 MBit, process switching to get ANDREAS.KLEMM.AK> andreas> *||**||**load balancing between 2 serial ANDREAS.KLEMM.AK> andreas> *||**|| ANDREAS.KLEMM.AK> andreas> *R2**R4 ANDREAS.KLEMM.AK> andreas> |**| ANDREAS.KLEMM.AK> andreas> *|10MBit**|10MBit ANDREAS.KLEMM.AK> andreas> *|e0**|e0 ANDREAS.KLEMM.AK> andreas> -----------+----------------- ANDREAS.KLEMM.AK> andreas> **| ANDREAS.KLEMM.AK> andreas> **|10 MBit ANDREAS.KLEMM.AK> andreas> **|e0 ANDREAS.KLEMM.AK> andreas> **Rdest ANDREAS.KLEMM.AK> andreas> ANDREAS.KLEMM.AK> andreas> All I read in the OSPF V2 RFC is, that OSPF creates one routing ANDREAS.KLEMM.AK> andreas> entry for the interface Rstart e0, that has to destinations ANDREAS.KLEMM.AK> andreas> R1 and R3 and that it does a load sharing between the two ways, ANDREAS.KLEMM.AK> andreas> but how ? ANDREAS.KLEMM.AK> ANDREAS.KLEMM.AK> OSPF V2 RFC only says "it can". Don't touch anything about how to ANDREAS.KLEMM.AK> share loads, because that is not routing protocol matter. ANDREAS.KLEMM.AK> ANDREAS.KLEMM.AK> How to share loads depends on how to lookup nexthop from routing ANDREAS.KLEMM.AK> table.So this is a issue of implementing routing table lookup function. ANDREAS.KLEMM.AK> ANDREAS.KLEMM.AK> andreas> How is the routing decision being done. What impact does load ANDREAS.KLEMM.AK> andreas> have on a routing decision and how often it is re-calculated ??? ANDREAS.KLEMM.AK> ANDREAS.KLEMM.AK> Again, this is a seperate matter, so the re-calculation will not be ANDREAS.KLEMM.AK> happen. ANDREAS.KLEMM.AK> ANDREAS.KLEMM.AK> andreas> When is perhaps only one route used between R1-R2 or R3-R4 ? ANDREAS.KLEMM.AK> andreas> ANDREAS.KLEMM.AK> andreas> The question is, in an redundant environment, how well does OSPF ANDREAS.KLEMM.AK> andreas> scale when it comes to the point of question, are and how well ANDREAS.KLEMM.AK> andreas> are the links shared. Do users get 2 x 4 = 8 MBit throughput or ANDREAS.KLEMM.AK> andreas> only 1 x 4 MBit ??? ANDREAS.KLEMM.AK> ANDREAS.KLEMM.AK> "How well" depends on algorithm to decide which way really to use. ANDREAS.KLEMM.AK> OSPF just makes multiple ways that can be used. ANDREAS.KLEMM.AK> ANDREAS.KLEMM.AK> *yasu ANDREAS.KLEMM.AK> ANDREAS.KLEMM.AK> To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-net" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19990528200257W.yasu>