badly worded question on subneting need help

pinkiaiii

so heres example q :

The network admin needs to address 7 lans,rip v1 is thew only routing protocol in use and subnet zero is not used,what is the maximum number of usable addresses can be supported on each lan,if organization is using one class c address block.

Now seeing and reading this question several times,conclusion ive made is that i need to find block that can support 7 ip per lan,thus ruling out 8 and closest being 16,but of course answer is wrong.Now i can check the answer but what gets me that i cant understand what im being asked for here

james43026

The question is asking that you define 7 subnets using a single class C address block, not asking you to define subnets with 7 hosts. What if the Class C block they give you is 192.168.1.0/24? How would you subnet that into 7 subnets? And how many hosts would be on each of those 7 subnets?

GDaines

pinkiaiii wrote: »

so heres example q :

The network admin needs to address 7 lans,rip v1 is thew only routing protocol in use and subnet zero is not used,what is the maximum number of usable addresses can be supported on each lan,if organization is using one class c address block.

Now seeing and reading this question several times,conclusion ive made is that i need to find block that can support 7 ip per lan,thus ruling out 8 and closest being 16,but of course answer is wrong.Now i can check the answer but what gets me that i cant understand what im being asked for here

Here's how I'm reading it:

You need to create address blocks for 7 LANs/subnets from one class C block. For Cisco that means a 255.255.255.0 mask, so you have 256 addresses to divide up. Subnet zero allows you to use the first and last blocks but it's been stated subnet zero is not being used so those blocks are out of the equation.

Using a block size of 32 only gives you 6 usable subnets of 30 addresses which doesn't meet the criteria specified for 7 LANs, so a block size of 16 gives you 14 usable addresses for each of up to 14 subnets.


So, what was the answer? Probably easier to understand the question once you know the answer.

OctalDump

RIPv1 doesn't support classless addressing, so it is going to depend very much on the topology here what would be practical. Static routing might be necessary. But it might be the case that you can only have one subnet in use. But I think without further information, this point can be ignored.

Possibly you can break up your /24 subnet (class c) into 8 /27 subnets. That would give you your 7 "lans" with subnet zero not used. A /27 would leave 5 bits for host addresses (minus 2 for the network address and broadcast address).

RIPv1 and no subnet zero, though. What is this? 1998?

GDaines

OctalDump wrote: »

RIPv1 doesn't support classless addressing

Which is fine as long as all LANs use the same subnet mask.

OctalDump

GDaines wrote: »

Which is fine as long as all LANs use the same subnet mask.

I'm not sure what you mean.

The situation is that you have one classful class C subnet, so you wouldn't be able to use RIPv1 to advertise routes to the /27 subnets if they were on different routers, since each router would then be advertising the same /24 subnet but only having one net connected. You could get it to work if you used one router and all the subnets were directly attached to that router, or if you configured static routing. Perhaps it could be something like a stubby area where the routers in that area are configured with static routes, and the gateway router is running RIPv1 on its 'outside' interface, since all subnets would need to go through the gateway, that would still work.

But I'm pretty sure that in this example it is presented as a red-herring and nothing more, since the routing protocol doesn't necessarily limit how you create your subnets, only how they are advertised to other routers.

Techytach

Last sentence is key for clarity. Previous stuff sets up scenario.

what is the maximum number of usable addresses can be supported on each lan,if organization is using one class c address block?

ripv1 means no vlsm, so they will be equal size. class c (out of 254) splitting into 7 means 32 (/27) (64 too large, 16 too small). no subnet 0. no network + no broadcast (-2). 32-2 = 30 usable addresses per lan.

GDaines

OctalDump wrote: »

RIPv1 doesn't support classless addressing

GDaines wrote: »

Which is fine as long as all LANs use the same subnet mask.

OctalDump wrote: »

I'm not sure what you mean.

Same as him Except that my understanding of ip subnet-zero was that the first and last subnets are not valid unless subnet zero is enabled, so I still stand by my answer that a block size of 32 allows for 8 subnets, minus the first and last because no subnet-zero leaving only 6. As we need addresses for 7 subnets we have to drop to a block size of 16 giving 14 usable host addresses per subnet.

http://www.cisco.com/c/en/us/support/docs/ip/dynamic-address-allocation-resolution/13711-40.html#subnetzero

Techytach wrote: »

Last sentence is key for clarity. Previous stuff sets up scenario. what is the maximum number of usable addresses can be supported on each lan,if organization is using one class c address block?

ripv1 means no vlsm, so they will be equal size. class c (out of 254) splitting into 7 means 32 (/27) (64 too large, 16 too small). no subnet 0, plus no broadcast (-2). 32-2 = 30 usable addresses per lan.

Techytach

Meant to say (sub)network + broadcast (-2) +32 = 30 not subnet 0 wrote that too quickly

GD - I think they are simply stating the subnet 0 isnt being used (the first subnet, meaning you need to plan for 8 not 7) But 30 should still be the answer (I think.) It's not that it was saying its disabled, where you would be right and have to plan for 2 extra. I take it as that they just aren't using the first address space. If it was disabled, I'd expect "disabled" as part of it.

Confusing terminology indeed. My 2c

OctalDump

Yeah, I think you are conflating VLSM and classless addressing. RIPv1 has no capacity to advertise the subnet size, equal size or other. All subnet sizes are based on the class of the network, in this case /24. RIPv1 is from 1988 and predates CIDR and VLSM by 5 years, so this lack of functionality is not surprising.

Techytach

I am no expert on RIPv1, and I'll go out on a limb and say you aren't either but while classful addressing doesn't advertise subnet masks so long as you are consistent you can indeed use a class c /27 for instance.

james43026

OctalDump wrote: »

Yeah, I think you are conflating VLSM and classless addressing. RIPv1 has no capacity to advertise the subnet size, equal size or other. All subnet sizes are based on the class of the network, in this case /24. RIPv1 is from 1988 and predates CIDR and VLSM by 5 years, so this lack of functionality is not surprising.

OctalDump is correct on this. RIPv1 does not support classless subnets at all. Furthermore the question is actually moot, as subnet zero would have to be used, as it's literally the only option. My original intention was to allow the OP to complete some subnetting, and then inform them that the question is actually an impossibility. But it appears we have already arrived there. What book, website, or materials did you get this question from?

Techytach

Here is a CCIP http://www.techexams.net/forums/ccna-ccent/15813-classeless-vs-vlsm-vs-cidr.html supporting my claim. I tried to figure out where I learned this but can't find source. This is the best I can link, sorry.

"This is incorrect, i can have a classful routing protocol with an address
151.78.1.1/25, in this case i must use /25 throughout my network when using a 151.78.x.x network address." edthelad

pinkiaiii

Techytach and James were correct on first response answer was 30, it basically gave numbers 8-16-30-32-64 as answers,what diged my head in simple question as you have class c mask and you need 7 subnets would be enough to know that one needs to work with block 32 thus mask 224 or /27 but whole 7 lans rip v1 subnet zero just messes up actual question where simple question as such makes you think why is it mentioning all useless info instead of asking for mask to fit 7 subnets into it .Since its clear one would need to remove -2 to get usable addresses,but what threw me off was saying 7 lans need to be addressed thus focus shifted towards fitting 7 ips into each subnet and choosing 16-since 8th block can fit only 6 addresses.Which wasnt anything like that in this case.

OctalDump

pinkiaiii wrote: »

it basically gave numbers 8-16-30-32-64 as answers

Well if those are the available options, then the question is full of red herrings. All you need to know is that the number of available host addresses will always be 2^n - 2, and only one of those options is 2^n - 2 - all the rest are 2^n. That kind of makes the question make more sense, since it becomes clear that they aren't really testing knowledge of RIPv1 or subnet zero or classes or anything else. They just want to test that you know that "the maximum number of usable addresses" is always 2^n - 2.

This skill is important part of test taking.

OctalDump

Techytach wrote: »

Here is a CCIP http://www.techexams.net/forums/ccna-ccent/15813-classeless-vs-vlsm-vs-cidr.html supporting my claim. I tried to figure out where I learned this but can't find source. This is the best I can link, sorry.

"This is incorrect, i can have a classful routing protocol with an address
151.78.1.1/25, in this case i must use /25 throughout my network when using a 151.78.x.x network address." edthelad

Read the RFCs. They are the final arbiter of truth for this. RIPv1 RFC is from 1988, CIDR (which introduces classless subnet masks) is from 1993. The RIPv2 RFC explicitly introduces a 'subnet field' to allow for classless subnet masks.

Subnet zero and the all-ones subnet are related concepts, but not the same thing. And obsolete for many years. Since the question only mentions one of these subnets, that is the only one you need to be concerned with.

james43026

OctalDump wrote: »

Read the RFCs. They are the final arbiter of truth for this. RIPv1 RFC is from 1988, CIDR (which introduces classless subnet masks) is from 1993. The RIPv2 RFC explicitly introduces a 'subnet field' to allow for classless subnet masks.

Subnet zero and the all-ones subnet are related concepts, but not the same thing. And obsolete for many years. Since the question only mentions one of these subnets, that is the only one you need to be concerned with.

I have to agree with OctalDump. Furthermore the post that you linked had incorrect information. Sure you could do that, but why the hell would you? Try reading over this it shows you exactly why you would never want to use a classless addressing scheme with a classful protocol. It just ends badly, and you have no way to control it. Anyone still in doubt needs to think about this, a classful protocol like RIPv1 will not send a subnet mask with it's routing updates, and makes an assumption on what subnet mask to use with any route / network it receives in an update based on the first octet of the IPv4 address. That means if you subnet an address and use RIPv1, then RIP doesn't even understand what subnetting is, nor does it care what you've done, it simply sends out IPv4 network addresses and allows neighbors the ability to assume what the subnet mask is.

"This is incorrect, i can have a classful routing protocol with an address
151.78.1.1/25, in this case i must use /25 throughout my network when using a 151.78.x.x network address." edthelad

This is literally completely false, unless you really want to have a bunch of routing loops and or black holes. In the example that edthelad gave, what if you create the network below?

151.78.1.1/25----RouterA

---

RouterB

---

RouterC

---

151.78.1.128/25

Router B now thinks it has two paths to reach the same network....which is 151.78.0.0/16, because the first octet is read as a class b network. Now what if the links leading either left or right from RouterB are all equal cost, now RouterB might even try to load balance said traffic VIA CEF. Essentially blackholeing part of your network traffic.

Techytach

You guys are being too practical. This is a thread about passing a cisco cert. They tend to be unrealistically technical. I was just giving relevant info about what I assume is a convoluted cisco practice question. I wasn't advocating it's use in any real life applications.

james43026

Cisco exams are all about being practical, according to their standards. And I doubt their standards would ever tell you to place a blackhole in your network.

Just for information purposes, I did find what edthelad was trying to elude to in his post. And as I mentioned, sure you could do this, but as far as scalability, you are talking about a nightmare. I suppose on a small network if you really wanted to do this, you could, I can't imagine the decision to do this, being anything other than money.



Routing Information Protocol version 1 (RIPv1) and Interior Gateway Routing Protocol (IGRP), common interior gateway protocols, cannot support VLSM because they do not send subnet information in their updates. Upon receiving an update packet, these classful routing protocols use one of the following methods to determine an address's network prefix:

• If the router receives information about a network, and if the receiving interface belongs to that same network, but on a different subnet, the router applies the subnet mask that is configured on the receiving interface.
  
• If the router receives information about a network address that is not the same as the one configured on the receiving interface, it applies the default, subnet mask (by class).
  
  Source: http://www.ciscopress.com/articles/article.asp?p=330807&seqNum=4
  
  

Techytach

james43026 wrote: »

Cisco exams are all about being practical, according to their standards. And I doubt their standards would ever tell you to place a blackhole in your network.

Scenarios/questions aren't telling you to do anything.

There were stupid questions on it. There was one in particular that was so ridiculous I wish I could talk about it. Fact is you have to pick the best answer available.

Hopefully the revised edition will be better.

GDaines

Techytach wrote: »

GD - I think they are simply stating the subnet 0 isnt being used (the first subnet, meaning you need to plan for 8 not 7) But 30 should still be the answer (I think.) It's not that it was saying its disabled, where you would be right and have to plan for 2 extra. I take it as that they just aren't using the first address space. If it was disabled, I'd expect "disabled" as part of it.

Confusing terminology indeed.

Yes you seem to be right, and the posts following yours clarify things a little further. I agree with the thread title in as much as it is definitely a badly worded question, just look at the fact that pinkiaiii first read it as needing 7 addresses per subnet, and I read no subnet zero as meaning the command 'no ip subnet-zero' had been used disabling the use of the first and last subnets. It's only once it had been pointed out a couple of times that I can see how the question meant it to be read, that and the fact that 30 is one of the possible answers whereas 14 isn't. Still not sure I'd have answered it right myself either though.

volfkhat

Thanks for this Post! I learn more everyday :]

First... yes, this question DEFINITELY SUCKS.

My brain got overwhelmed and decided this was just a question testing your VLSM knowledge.
Hence, the answer is clearly something with /27.... which gives your 8 subnets.
Then, throw in the fact they said to skip the first subnet (ZERO), and that leaves you with the 7 subnets (the same # asked for in the question).
Voila!

I have no practical experience with RIPv1 (nor Do i want to learn).
But your guys are 100% correct about the lack of vlsm capabilities (i forgot).
So actually... my initial answer... is WRONG.

EDIT:
After giving it some thought... i have a better answer:
If you review the exam objectives... you will see that Cisco removed 'RIP' topics from the ICND1 & 2 exams.

Thus, Invalid Question ;]

dontstop

OctalDump wrote: »

RIPv1 and no subnet zero, though. What is this? 1998?

1 10 char

OctalDump

Read the question, and then read the answers. Figure out what you can exclude. In this case you could exclude all but one answer just knowing that the question was asking for number of available host addresses. It's not uncommon for these questions to include "distractors" irrelevant information that's there to confuse you (and can actually confuse you more if you know more). So figuring out what parts of a question (or what answers) you can ignore can be a very handy skill.

There is a chance you could see something similar, although they tend to put the irrelevant stuff in the answers more than the question. The v3 ICND exams do include RIPv2, so that is something to be aware of. They will expect you to know, for instance, that RIPv2 is classful and what that means.