|
Home > Archive > alt.certification.network-plus > December 2003 > Split Horizon and Poison Reverse
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
| Author |
Split Horizon and Poison Reverse
|
|
| geomax 2003-12-18, 8:24 pm |
| I'm doing some review of distance vector routing, and my study materials do
a very poor job of explaining 2 concepts: Split Horizon and Poison Reverse.
Can some point me to some material that explains these clearly? I get count
to infinity loops, and how SH and PR prevent looping. It's just the details
how how they do it is causing me some confusion.
Thanks!
| |
| SIR-BUZZ 2003-12-19, 2:25 am |
| This Might give you a heads up.
http://www.cisco.com/pcgi-bin/searc...Connection&lang
uage=en&country=US&accessLevel=Guest&siteToSearch=cisco.com&x=9&y=5
On Fri, 19 Dec 2003 01:11:38 GMT, "geomax" <gw091@yahoo.com> wrote:
>I'm doing some review of distance vector routing, and my study materials do
>a very poor job of explaining 2 concepts: Split Horizon and Poison Reverse.
>
>Can some point me to some material that explains these clearly? I get count
>to infinity loops, and how SH and PR prevent looping. It's just the details
>how how they do it is causing me some confusion.
>
>Thanks!
>
| |
| Charles M. Kozierok 2003-12-19, 2:25 pm |
| In article <efsEb.1596$J14.91@nwrdny01.gnilink.net>,
geomax <gw091@yahoo.com> wrote:
} I'm doing some review of distance vector routing, and my study materials do
} a very poor job of explaining 2 concepts: Split Horizon and Poison Reverse.
}
} Can some point me to some material that explains these clearly? I get count
} to infinity loops, and how SH and PR prevent looping. It's just the details
} how how they do it is causing me some confusion.
Imagine a simple internetwork as follows:
Network 1 -- Router A -- Network 2 -- Router B
RB can only reach N1 through RA. Counting to infinity occurs when the
link between RA and N1 goes down, and RA later receives a route message
from RB that tricks RA into thinking it can access N1 through RB.
Split horizon institutes the rule that a router cannot advertise a route
back to the router from which it initially learned it. So, since RB
learned of the route to N1 through RA, RB is not allowed to advertise N1
when sending on N2. This prevents RA from being fooled by RB's
advertisement.
Split horizon with poisoned reverse adds an extra form of "insurance".
Instead of RB not advertising back to RA, it sends an advertisement to
RA with a cost metric of 16 (RIP infinity). This explicitly tells RA
"you cannot get to N1 through RB" and prevents the problem entirely.
Hope that helped. I explain this in much more detail (with illustration)
in my upcoming TCP/IP Guide, which should be available in 2 to 3 weeks.
peace,
-*-
charles
| |
|
|
"Charles M. Kozierok" <ixlu@PCGuide.com> wrote in message
news:brvde1$390$1@reader2.panix.com...
> In article <efsEb.1596$J14.91@nwrdny01.gnilink.net>,
> geomax <gw091@yahoo.com> wrote:
> } I'm doing some review of distance vector routing, and my study materials
do
> } a very poor job of explaining 2 concepts: Split Horizon and Poison
Reverse.
> }
> } Can some point me to some material that explains these clearly? I get
count
> } to infinity loops, and how SH and PR prevent looping. It's just the
details
> } how how they do it is causing me some confusion.
>
> Imagine a simple internetwork as follows:
>
> Network 1 -- Router A -- Network 2 -- Router B
>
> RB can only reach N1 through RA. Counting to infinity occurs when the
> link between RA and N1 goes down, and RA later receives a route message
> from RB that tricks RA into thinking it can access N1 through RB.
>
> Split horizon institutes the rule that a router cannot advertise a route
> back to the router from which it initially learned it. So, since RB
> learned of the route to N1 through RA, RB is not allowed to advertise N1
> when sending on N2. This prevents RA from being fooled by RB's
> advertisement.
>
> Split horizon with poisoned reverse adds an extra form of "insurance".
> Instead of RB not advertising back to RA, it sends an advertisement to
> RA with a cost metric of 16 (RIP infinity). This explicitly tells RA
> "you cannot get to N1 through RB" and prevents the problem entirely.
>
> Hope that helped. I explain this in much more detail (with illustration)
> in my upcoming TCP/IP Guide, which should be available in 2 to 3 weeks.
>
> peace,
>
> -*-
> charles
Is this included in the N+ exam material?
| |
|
|
"Jim" <James@yahoo.com> wrote in message
news:UpNEb.217295$Ec1.7788635@bgtnsc05-news.ops.worldnet.att.net...
>
>
> Is this included in the N+ exam material?
>
No. My test had a few questions about routing protocols but nothing beyond
RIP and OSPF. You should know the difference between routing and routable
protocols, etc.
| |
| geomax 2003-12-21, 11:25 am |
| Hi Charles,
Thanks for the helpful reply. Please keep me informed about your guide. I
would be interested in a copy!
G
> Imagine a simple internetwork as follows:
>
> Network 1 -- Router A -- Network 2 -- Router B
>
> RB can only reach N1 through RA. Counting to infinity occurs when the
> link between RA and N1 goes down, and RA later receives a route message
> from RB that tricks RA into thinking it can access N1 through RB.
>
> Split horizon institutes the rule that a router cannot advertise a route
> back to the router from which it initially learned it. So, since RB
> learned of the route to N1 through RA, RB is not allowed to advertise N1
> when sending on N2. This prevents RA from being fooled by RB's
> advertisement.
>
> Split horizon with poisoned reverse adds an extra form of "insurance".
> Instead of RB not advertising back to RA, it sends an advertisement to
> RA with a cost metric of 16 (RIP infinity). This explicitly tells RA
> "you cannot get to N1 through RB" and prevents the problem entirely.
>
> Hope that helped. I explain this in much more detail (with illustration)
> in my upcoming TCP/IP Guide, which should be available in 2 to 3 weeks.
>
> peace,
>
> -*-
> charles
| |
| geomax 2003-12-21, 12:25 pm |
| Charles, question for you:
This may sound like a dumb question, but with Split Horizon, even though
they can't *advertise* back to the router from which it initially learned of
a route, they still added the route to their routing table, so RB can still
route traffic to N1 via RA, right??
Split Horizon/Poison Reverse are all about ellimianting false advertising -
but don't change the way the routers actually route traffic, right?
Poison Reverse? In this case, since RA is told it can't get to N1 via RB, it
has to find another way? I'm assuming then, given your example, that there
would have to be other routers/routes that RA can send traffic to to get to
N1?
Is it a safe assumption to say that you would use Split Horizon in a small
network, and Poison Reverse in a more complex meshed environment?
I think I'm confusing how routers update each other and how they end up
actually routing.
"Charles M. Kozierok" <ixlu@PCGuide.com> wrote in message
news:brvde1$390$1@reader2.panix.com...
> In article <efsEb.1596$J14.91@nwrdny01.gnilink.net>,
> geomax <gw091@yahoo.com> wrote:
> } I'm doing some review of distance vector routing, and my study materials
do
> } a very poor job of explaining 2 concepts: Split Horizon and Poison
Reverse.
> }
> } Can some point me to some material that explains these clearly? I get
count
> } to infinity loops, and how SH and PR prevent looping. It's just the
details
> } how how they do it is causing me some confusion.
>
> Imagine a simple internetwork as follows:
>
> Network 1 -- Router A -- Network 2 -- Router B
>
> RB can only reach N1 through RA. Counting to infinity occurs when the
> link between RA and N1 goes down, and RA later receives a route message
> from RB that tricks RA into thinking it can access N1 through RB.
>
> Split horizon institutes the rule that a router cannot advertise a route
> back to the router from which it initially learned it. So, since RB
> learned of the route to N1 through RA, RB is not allowed to advertise N1
> when sending on N2. This prevents RA from being fooled by RB's
> advertisement.
>
> Split horizon with poisoned reverse adds an extra form of "insurance".
> Instead of RB not advertising back to RA, it sends an advertisement to
> RA with a cost metric of 16 (RIP infinity). This explicitly tells RA
> "you cannot get to N1 through RB" and prevents the problem entirely.
>
> Hope that helped. I explain this in much more detail (with illustration)
> in my upcoming TCP/IP Guide, which should be available in 2 to 3 weeks.
>
> peace,
>
> -*-
> charles
| |
| Frederic 2003-12-22, 10:25 am |
| Split Horizon and Route poisining are related to CCNA material.
This question in fact should has been asked to alt.certification.cisco
"Split Horizon: Another solution to the routing loop problem is called split
horizon. This reduces incorrect routing information and routing overhead in
a distance-vector network by enforcing the rule that information cannot be
sent back in the direction from which it was received. In other words, the
routing protocol differentiates which interface a network route was learned
on and then will not advertise that route back out that same interface. This
would have prevented Router A from sending the updated information it
received from Router B back to Router B.
Route Poisoning: Another way to avoid problems caused by inconsistent
updates and stop network loops is route poisoning. For example, when Network
5 goes down, Router E initiates route poisoning by entering a table entry
for Network 5 as 16, or unreachable (sometimes referred to as infinite). By
this poisoning of the route to Netowkr 5, Router C is not susceptible to
incorrect updates about the route to Network 5. When Router C receives a
router poisoning from Router E, it sends an update, called a poison reverse,
back to Router E. This makes sure all routes on the segment have received
the poisoned route information.
Route poisoning and split horizon create a much more resilient and
dependable distance-vector network than we'd have otherwise and stop network
loops."
Sybex CCNA Study Guide Exam Deluxe Edition
--
Frederic
MCP, IT Project+, i-Net+, CIWA, A+
member of: CompTIA-ITPRO, HDI, IETF
------------------------------------------------------
http://fredsfastcram.netfirms.com
------------------------------------------------------
"geomax" <gw091@yahoo.com> wrote in message
news:rJjFb.7182$xh2.1338@nwrdny02.gnilink.net...
> Charles, question for you:
>
> This may sound like a dumb question, but with Split Horizon, even though
> they can't *advertise* back to the router from which it initially learned
of
> a route, they still added the route to their routing table, so RB can
still
> route traffic to N1 via RA, right??
>
> Split Horizon/Poison Reverse are all about ellimianting false
advertising -
> but don't change the way the routers actually route traffic, right?
>
> Poison Reverse? In this case, since RA is told it can't get to N1 via RB,
it
> has to find another way? I'm assuming then, given your example, that there
> would have to be other routers/routes that RA can send traffic to to get
to
> N1?
>
> Is it a safe assumption to say that you would use Split Horizon in a small
> network, and Poison Reverse in a more complex meshed environment?
>
> I think I'm confusing how routers update each other and how they end up
> actually routing.
>
>
>
> "Charles M. Kozierok" <ixlu@PCGuide.com> wrote in message
> news:brvde1$390$1@reader2.panix.com...
materials[color=blue]
> do
> Reverse.
> count
> details
>
>
| |
| Charles M. Kozierok 2003-12-22, 12:25 pm |
| In article <rJjFb.7182$xh2.1338@nwrdny02.gnilink.net>,
geomax <gw091@yahoo.com> wrote:
} This may sound like a dumb question, but with Split Horizon, even though
} they can't *advertise* back to the router from which it initially learned of
} a route, they still added the route to their routing table, so RB can still
} route traffic to N1 via RA, right??
Correct. It only impacts the advertising of route information.
} Split Horizon/Poison Reverse are all about ellimianting false advertising -
} but don't change the way the routers actually route traffic, right?
Right.
} Poison Reverse? In this case, since RA is told it can't get to N1 via RB, it
} has to find another way? I'm assuming then, given your example, that there
} would have to be other routers/routes that RA can send traffic to to get to
} N1?
In that simplified example, there would be no way for anyone to reach N1
until the link between N1 and RA was repaired.
} Is it a safe assumption to say that you would use Split Horizon in a small
} network, and Poison Reverse in a more complex meshed environment?
I am not really sure about this. Poison reverse is just a "safer"
version of split horizon and I believe it is widely used on
internetworks of varying size.
peace,
-*-
charles
| |
| geomax 2003-12-22, 8:25 pm |
|
> Split Horizon and Route poisining are related to CCNA material.
> This question in fact should has been asked to alt.certification.cisco
Sorry 'bout that, but these concepts did appear in my courseware for
Network+, and since I plan to go onto the CCNA, I just wanted to get it
straight now.
> "Split Horizon: Another solution to the routing loop problem is called
split
> horizon. This reduces incorrect routing information and routing overhead
in
> a distance-vector network by enforcing the rule that information cannot be
> sent back in the direction from which it was received. In other words, the
> routing protocol differentiates which interface a network route was
learned
> on and then will not advertise that route back out that same interface.
This
> would have prevented Router A from sending the updated information it
> received from Router B back to Router B.
> Route Poisoning: Another way to avoid problems caused by inconsistent
> updates and stop network loops is route poisoning. For example, when
Network
> 5 goes down, Router E initiates route poisoning by entering a table entry
> for Network 5 as 16, or unreachable (sometimes referred to as infinite).
By
> this poisoning of the route to Netowkr 5, Router C is not susceptible to
> incorrect updates about the route to Network 5. When Router C receives a
> router poisoning from Router E, it sends an update, called a poison
reverse,
> back to Router E. This makes sure all routes on the segment have received
> the poisoned route information.
> Route poisoning and split horizon create a much more resilient and
> dependable distance-vector network than we'd have otherwise and stop
network
> loops."
> Sybex CCNA Study Guide Exam Deluxe Edition
Thats great info - thanks!
| |
| Frederic 2003-12-22, 11:25 pm |
| Network+ objectives for Routing are:
1.6 Identify the purpose, features, and functions of the following network
components:
a.. Hubs
b.. Switches
c.. Bridges
d.. Routers
e.. Gateways
f.. CSU/DSU
g.. Network Interface Cards/ISDN adapters/system area network cards
h.. Wireless access points
i.. Modems
http://www.comptia.org/certificatio...ves-domain1.asp
2.3 Differentiate between the following network protocols in terms of
routing, addressing schemes, interoperability and naming conventions:
a.. TCP/IP
b.. IPX/SPX
c.. NetBEUI
d.. AppleTalk
2.4 Identify the OSI layers at which the following network components
operate:
a.. Hubs
b.. Switches
c.. Bridges
d.. Routers
e.. Network Interface Cards
http://www.comptia.org/certificatio...ves-domain2.asp
Split Horizon and Poison Reverse are way beyond Network+ objectives.
It's not bad that it is included in a Network+ book. It's just an extra
knowledge.
--
Frederic
MCP, IT Project+, i-Net+, CIWA, A+
member of: CompTIA-ITPRO, HDI, IETF
------------------------------------------------------
http://fredsfastcram.netfirms.com
------------------------------------------------------
"geomax" <gw091@yahoo.com> wrote in message
news:ubMFb.37$834.30@nwrdny01.gnilink.net...
>
>
> Sorry 'bout that, but these concepts did appear in my courseware for
> Network+, and since I plan to go onto the CCNA, I just wanted to get it
> straight now.
>
> split
> in
be[color=blue]
the[color=blue]
> learned
> This
> Network
entry[color=blue]
> By
> reverse,
received[color=blue]
> network
>
> Thats great info - thanks!
>
>
| |
| geomax 2003-12-22, 11:25 pm |
| I was more interested in satisfying my curiosity and interest. Extra
knowledge is a good thing.
You've got some great Network+ info on your site. Seems like the i-Net+
notes also have some very useful information. Not all applies, but there's
some overlap.
Thanks again!
"Frederic" <frederic.l@softhome.net> wrote in message
news:auOFb.54176$NT6.1257248@weber.videotron.net...
> Network+ objectives for Routing are:
>
> 1.6 Identify the purpose, features, and functions of the following network
> components:
> a.. Hubs
> b.. Switches
> c.. Bridges
> d.. Routers
> e.. Gateways
> f.. CSU/DSU
> g.. Network Interface Cards/ISDN adapters/system area network cards
> h.. Wireless access points
> i.. Modems
>
http://www.comptia.org/certificatio...ves-domain1.asp
>
> 2.3 Differentiate between the following network protocols in terms of
> routing, addressing schemes, interoperability and naming conventions:
> a.. TCP/IP
> b.. IPX/SPX
> c.. NetBEUI
> d.. AppleTalk
> 2.4 Identify the OSI layers at which the following network components
> operate:
> a.. Hubs
> b.. Switches
> c.. Bridges
> d.. Routers
> e.. Network Interface Cards
>
http://www.comptia.org/certificatio...ves-domain2.asp
>
> Split Horizon and Poison Reverse are way beyond Network+ objectives.
> It's not bad that it is included in a Network+ book. It's just an extra
> knowledge.
>
> --
> Frederic
> MCP, IT Project+, i-Net+, CIWA, A+
> member of: CompTIA-ITPRO, HDI, IETF
> ------------------------------------------------------
> http://fredsfastcram.netfirms.com
> ------------------------------------------------------
>
> "geomax" <gw091@yahoo.com> wrote in message
> news:ubMFb.37$834.30@nwrdny01.gnilink.net...
overhead[color=blue]
cannot[color=blue]
> be
> the
interface.[color=blue]
> entry
infinite).[color=blue]
to[color=blue]
a[color=blue]
> received
>
>
|
|
|
|
|