Route summarization questions

rick1960

I doubt anyone is paying attention to this anymore, but, according to Cisco CCNP, BSCI, all of these answers over summarize, and results in flapping and over utilization of processor and memory.

kryolla

rick1960 wrote: »

I doubt anyone is paying attention to this anymore, but, according to Cisco CCNP, BSCI, all of these answers over summarize, and results in flapping and over utilization of processor and memory.

What are you talking about

kryolla

bakes00 wrote: »

Hi guys,

Ok, I have been practicing summarizing, but only in the forward sense.

I found an example question that reads:

- If a routers has networks connected to it that are summarized as 192.168.176.0/21, which packets will be sent to this router.

The question then lists 5 IP address options to select from.

Can anyone please show me how to do this process in the reverse to find the answer to the question?

Thanks in advance.

176 in binary is
10110000 with a mask of /21 so the third octet is
11111 so they have the first 5 bits in common according to what summarization is so the range what be to convert the last 3 bits of the IP address to all 1

so you end of up with 176 + 3 bits which is 7 the range is 176-183

phoeneous

kryolla wrote: »

so the third octet is 11111 so they have the first 5 bits in common according to what summarization is

You lost me right there. Where do you get 11111 from?

kryolla

phoeneous wrote: »

You lost me right there. Where do you get 11111 from?

the mask is /21 so in the third octet the first 5 bits are 1's 8 for the 1st octet and another 8 for the second octet and 5 in the third to equal /21 the third octet is the interesting octet so that is where all the calculations are done

192.168.176.0/24
to
192.168.183.0/24

phoeneous

kryolla wrote: »

the mask is /21 so in the third octet the first 5 bits are 1's 8 for the 1st octet and another 8 for the second octet and 5 in the third to equal /21 the third octet is the interesting octet so that is where all the calculations are done

192.168.176.0/24
to
192.168.183.0/24

I know that, I was just confused with what you meant by "in common according to what summarization is".

kryolla

phoeneous wrote: »

I know that, I was just confused with what you meant by "in common according to what summarization is".

summarization is all about finding the most bits in network portion of the IP addresses in common and that will be your mask so if you have 21 bits out of 24 that is common then your mask is /21, the other NETWORK bits that are not common when you make them all 0's will be you summarization address and then to find the range just invert the uncommon bits to all 1's.

chargen

mikearama wrote: »

Now, slightly tougher (from our actual network):
10.22.178.0 /23
10.22.180.0 /23
10.22.182.0 /23
10.22.184.0 /23
10.22.186.0 /23
10.22.188.0 /23
10.22.190.0 /23
... what's the summary address / mask?

Have fun,
mike

I stumbled across this while searching for CCNP BSCI summarization practice questions. Doesn't the answer of 10.22.176.0/20 oversummarize?

The answer I came up with which I think avoids over-summarization is:

10.22.178.0/23
10.22.180.0/22
10.22.184.0/21

Correct or am I missing something?

mwgood

chargen wrote: »

I stumbled across this while searching for CCNP BSCI summarization practice questions. Doesn't the answer of 10.22.176.0/20 oversummarize?

Yes.

chargen wrote: »

The answer I came up with which I think avoids over-summarization is:

10.22.178.0/23
10.22.180.0/22
10.22.184.0/21

Correct or am I missing something?

You are not missing anything. Your answer is correct, 10.22.176.0 /20 is not correct since it contains addresses outside of the originally specified address space.

censo

chargen wrote: »

I stumbled across this while searching for CCNP BSCI summarization practice questions. Doesn't the answer of 10.22.176.0/20 oversummarize?

The answer I came up with which I think avoids over-summarization is:

10.22.178.0/23
10.22.180.0/22
10.22.184.0/21

Correct or am I missing something?

Very ... VERY good observation. With that said, do you not see a similar problem in this proposed solution?

mikearama wrote: »

Those networks are remote, but that doesn't mean the entire 172.16.0.0 /16 is remote.

This kinda scheme just takes a bit of planning, since the initial ip scheme wasn't well planned. If it was, then local and remote networks would be in completely different core subnets.

You can group most of the subnets with 172.16.64.0 /18, which would cover 172.16.64.0 through 172.16.127.254

Now you have a choice... either leave the other three lines in as they are, or summarize again. Personally, I'd summarize again with 172.16.128.0 /18, which covers 172.16.128.0 through 172.16.191.254

I just wanted to get you techies thinking beyond the typical, easy, everyday subnetting question where everything fits into one summary address.

HTH,
Mike

The scenario is:

172.16.64.0 /20
172.16.80.0 /20
172.16.96.0 /20
172.16.112.0 /20
172.16.128.0 /20
172.16.144.0 /20
172.16.160.0 /20

By my calculations, that last subnet ends at 172.16.175.255. A summary route address of 172.16.128.0 /18 over summarizes as well.

What I originally came up with is the following:

172.16.64.0 /18
172.16.128.0 /19
172.16.160.0/20 (left as-is)

Wouldn't you agree this solution is the best case scenario where networks that don't belong to this scheme aren't inadvertently included? (e.g. " Personally, I'd summarize again with 172.16.128.0 /18, which covers 172.16.128.0 through 172.16.191.254")

Wave breaker

bakes00 wrote: »

Hi guys,

Please help me, I seem to have a fundamental misunderstanding here.

Everytime I do the route summary for these addresses I come up with:
192.168.0.0/20

How do you arrive at 192.168.8.0/21?

192.168.15.0 = 192.168.00001111.00000000

Only the last 4 bits are ever changed to reach 15, so the first 4 are all common among the given network addresses?


I guess I'm missing the point, but I thought I understood it.

Hope someone has the patience to help me understand how everyone arrived at the other answer.

hi bakes

you are confuse its simple

192.168.10.0 /24
192.168.11.0 /24
192.168.12.0 /24
192.168.13.0 /24
192.168.14.0 /24
192.168.15.0 /24


128 64 32 16 8 4 2 1

0 0 0 0 1 0 1 0 for 10.0
0 0 0 0 1 0 1 1 for 11.0
0 0 0 0 1 1 0 0 for 12.0
0 0 0 0 1 1 0 1 for 13.0
0 0 0 0 1 1 1 0 for 14.0
0 0 0 0 1 1 0 1 for 15.0

now how many bit are same 5 now for the first two oct it is 8+8+5 = 21

192.168.10.0/21

Wave breaker

bakes00 wrote: »

Hi guys,

Please help me, I seem to have a fundamental misunderstanding here.

Everytime I do the route summary for these addresses I come up with:
192.168.0.0/20

How do you arrive at 192.168.8.0/21?

192.168.15.0 = 192.168.00001111.00000000

Only the last 4 bits are ever changed to reach 15, so the first 4 are all common among the given network addresses?


I guess I'm missing the point, but I thought I understood it.

Hope someone has the patience to help me understand how everyone arrived at the other answer.

hi bakes

you are confuse its simple

192.168.10.0 /24
192.168.11.0 /24
192.168.12.0 /24
192.168.13.0 /24
192.168.14.0 /24
192.168.15.0 /24


128 64 32 16 8 4 2 1

0 0 0 0 1 0 1 0 for 10.0
0 0 0 0 1 0 1 1 for 11.0
0 0 0 0 1 1 0 0 for 12.0
0 0 0 0 1 1 0 1 for 13.0
0 0 0 0 1 1 1 0 for 14.0
0 0 0 0 1 1 0 1 for 15.0

now how many bit are same 5 now for the first two oct it is 8+8 and 3rd one is 5 so 8+8+5=21

192.168.10.0/21

Stilgar1973

This thread is like 100 years old.
But still very, very pertinent.
You know, I did a Google search for 'route summarization quiz' and didn't get ****? For someone self studying that needs a quiz to help solidify this concept, there isn't much out there except for this thread.

Thanks.

popquizkid95

Sorry to dig up such an old thread, but Laz mentioned that; alternative to giving you a list of IP addresses to summarize, the exam may ask you to do the reverse; what would be the network range for the route summary?

Example: 192.168.100.0 /22

I can't quite wrap my brain around how to do this. I use Chris Bryants method of fundamentally breaking down addresses using binary, so no tricks here.

pinkiaiii

some headache when it comes to counting but hope this will solve your trouble: mask is 252 =3 bits but all of them 100-103 are 1 thus block is 4 where difference would occur.

192.168.100.0 /22

255.255.252.0

100

0110.0100 100
0110.0101 101
0110.0110 102
0110.0111 103
0101.1000 104 <not valid

Not sure is that the answer your looking since newbie at summarization,but seems if would be them 4 addresses.

so this would be summary routes
192.168.100.0 /22 6 bits
192.168.101.0 /22 6 bits
192.168.102.0 /22 6 bits
192.168.103.0 /22 6 bits
192.168.104.0 /21 5 bits

popquizkid95

Ah yes, I understand that.. Thanks for the response. I may have another practice question to post soon, just to make sure I've got it.

james43026

What I've noticed in this thread, is that everyone seems to want their answers to cover far more subnets than what is needed for the summarization. You will never want to do this in a production environment. You would use the supernet that is the closest while still being too small to cover the entire range. And then just propagate the rest of the routes individually, or as a second supernet.

Example, you have the following networks that you need to summarize
192.168.1.0/29
192.168.1.8/28
192.168.1.24/28

Which means you need to summarize the following networks 192.168.1.0/29, 192.168.1.8/28, and 192.16.1.24/28, covering from 192.168.1.0 - 192.168.1.39. If you use 192.168.1.0/26, you would be adveritising 192.168.1.0 - 192.168.1.63, the problem is the fact that now you are advertising networks that don't exist (192.168.1.41 - 192.168.1.63), or they do exist somewhere else on the network, and now you just created a blackhole.

The correct answer would be to use 192.168.1.0/27 to cover 192.168.1.0 - 192.168.1.31, and then 192.192.1.32/29 to cover 192.168.1.32 - 192.168.1.39. I tried to simply this example as much as possible. The point is that you never want to advertise a supernet to cover addresses that aren't actually reachable by the router that is advertising them. This is the biggest mistake that people make when they are manually creating summarizations / supernets to advertise through the network. This leads to a blackhole.

pinkiaiii

got a bit lost there James but wouldnt this section here:The purpose of the Null0 summary route is to prevent routing loops for destinations that are included in the summary, but do not actually exist in the routing table < given that one would be using routing protocol to summarize routes thus even using manual summarization router would just discard any packets destined to non existent network ?

as for networks you posted mask alone would indicate that there should be more then one summary route,only not sure how to play with different masks,since by your example one would need just to combine same masks into summary routes,but you did it other way around and split it into combinging 1.0-with 1.8,instead of doing to 1.0 to 1.7 and another for 1.8 to 1.24 so cant see where is that mark for cut of point of getting /27 when putting it into binary.

29
0000.0001.0000.0000 1.0
28
0000.0001.0000.1000 1.8
28
0000.0001.0001.0100 1.24

so question remains black hole <would it discard any packages or what implications it would have ?

james43026

The Null0 interface is only present if configured correctly, that means for a dynamic routing protocol like EIGRP, OSPF, and BGP you would need to use the correct summarization commands or risk creating a routing loop and or a blackhole. Or you can statically configure a route to Null0 as well. You have to remember that the Null0 interface does only exist on the device that is advertising the summary route, that means any device that receives that summary route will still think that the device that originated the summary route can still reach the networks that are in the summary route, which means you would be wasting bandwidth, buffer space, adding latency to other traffic that are part of the same queues, for absolutely no good reason. That Null0 interface is a blackhole, meaning any packets routed to it, are just dropped, usually an ICMP net unreachable message is sent back to the node that originated the traffic. This should be avoided anytime that you can. Not only that, imagine the implications of setting up two different summary routes, that overlap, onto two different routers. The implication could be that you have a router load balancing to both originating routers, if you happen to be using a routing protocol that supports ECMP, which is automatically used in OSPF and EIGRP. Now say one of those summary routes sent out was legit, and the other was just a lazy summary route that someone configured. Now some of your routers that are using ECMP are load balancing to both routes to that same network, only one of those routes leads to a router that is using a Null0 interface, and the other part of the traffic is going to the legit route. You now have a blackhole, packets that are mysteriously dissappearing in your network.






192.168.1.0/29
Network address: 192.168.1.0
Host range: 192.168.1.1 - 192.168.1.6
Broadcast address: 192.168.1.7


192.168.1.8/28
Network address:192.168.1.8
Host range:192.168.1.9 - 192.168.1.22
Broadcast address:192.168.1.23


192.168.1.24/28
Network address:192.168.1.24
Host range:192.168.1.25 - 192.168.1.38
Broadcast address:192.168.1.39




When you are trying to figure out the summary address, in some cases you can't simply take the binary form of the IP address, and figure out what matches all of them, if my example it shouldn't be done because you would end up with this.


192.168.1.0: 00000000 last octect
192.168.1.8: 00001000 last octect
192.168.1.24:00011000 last octect


192.168.1.0/27 is what you would get from summarizing by the traditional method of convert to binary and find the last common bit amongst the addresses. That would mean a range of 192.168.1.0 - 192.168.1.31, this doesn't cover all of the networks that we need. This happens mainly because I'm using VLSM, which I did on purpose to show that the way summarization is taught is generally not correct. This is a shortcut that should be avoided once you get into a real production network, that will most likely be using VLSM.


If I had used non VLSM networks, something like


192.168.1.0/29: 00000000
192.168.1.8/29: 00001000
192.168.1.16/29:00010000


Summary is 192.168.1.0/27
Range: 192.168.1.0 - 192.168.1.32


Which would properly cover the networks needed, nice and neatly. Unfortunately most people that try to write articles about summarization / supernetting, and make videos about it, aren't telling people the proper practices, they are only giving textbook examples, and unfortunately that doesn't cut it for the real world.

pinkiaiii

Ok that was very clear and fair answer to Null0 and blackhole since far enough to know balancing now-so see how when in place and in action packets would go missing,when theres BH on one side thus either loss of packages or depending on transport layer tcp upd thus causing a lot of requests or getting errors in data transmitted.


But still didnt grasp how you would know when having different vlsm networks which way to summarize them ?

since in your example you gave just 1.0 /28 1.8/29 1.24/29
thus confussion comes not knowing how did you choose to summarize 1.0 with 1.8 when they have separate masks,instead of summarizing same masks and leaving 1.0 to 1.7.and doing separate one for /29 network.

thus not sure how different masks are used to combine such networks,btw i know this is out of ccna scope,but really interested to know while at small network this would be minor issue but as said on ISP line or along the lines of BGP it would be major impact.

james43026

For the sake of brevity, I'm going to give you a link to a great article.

pinkiaiii

So the way to do it,going by that article would be very close as another user here asked for reversing summary route instead of focusing on counting bits you focus on block that addresses fall thus,block size being marker as to your subnet mask and then getting your mask out of it-like in the articles example networks 1.0,2.0,3.64,4.128,4.192 =thus its not the mask one would need to look but third octets block size which falls onto 4th thus meaning 3 bits left out.
Quite strange given that nobody covers such topic in ccna given that it most likely would be beneficial to know.

wlkankota

mikearama wrote: »

Those networks are remote, but that doesn't mean the entire 172.16.0.0 /16 is remote.

This kinda scheme just takes a bit of planning, since the initial ip scheme wasn't well planned. If it was, then local and remote networks would be in completely different core subnets.

You can group most of the subnets with 172.16.64.0 /18, which would cover 172.16.64.0 through 172.16.127.254

Now you have a choice... either leave the other three lines in as they are, or summarize again. Personally, I'd summarize again with 172.16.128.0 /18, which covers 172.16.128.0 through 172.16.191.254

I just wanted to get you techies thinking beyond the typical, easy, everyday subnetting question where everything fits into one summary address.

HTH,
Mike

Anyone please recommend me some books that will help to this answer .