Variable Length Subnet Mask 1

[rclough]

Hi, hopefully this is the right forum for my question.

I am beginning my first year of teaching Cisco I and about to begin teaching about subnetting. Personally I hated this part of the course and I really get confused with the VLSM. Does any one have any advice or sites they could recommend for me to use? My class has had no experience with Cisco so they are all new learners and really struggling so I want to make it as simple as possible.

Thanks for the help...

 

[brianinms]

This is one of the best subnetting sites I have found. I use it to teach the new engineers.

http://www.learntosubnet.com/

 

[vipergg]

Brian is correct it is the best free site , and it is really a course so you can sit back and relax while they listen to this course which is probably a good 2-4 hours in length .

 

[jneiberger]

Learn to subnet in binary first. Then you can actually see what you're doing. Once you understand it you can move to doing it in decimal.

I like to think of subnetting as dividing up a pie into smaller pieces of variable size.

Let's say you've been assigned a /22 prefix. That's a lot of addresses in a single subnet, so you're going to want to break it down and spread it around your network as needed. You whole "pie" is a /22. If you wanted to divide it into two halves, you add one extra bit to your subnet mask and end up with two /23 prefixes. Each time you add an extra bit of subnet mask, you divide the prefix into two.

Two /23 prefixes is still a lot of addresses per subnet, so let's further divide those two pieces into halves. Now you have four /24 subnets. That's pretty good, but what if you have a couple of subnetworks on your network that just aren't going to need that many addresses? Take one of those /24s and divide it even further.

You could divide it into halves and end up with two /25s or go even further and divide it into four /26s. If you did that, you've now subnetted your /22 into three /24s and four /26s.

Is that helping at all?

 

[jneiberger]

Continuing my pie analogy, I would recommend that you actually draw out the pie as you subnet the network. I'm fairly good at subnetting but I still draw the pie and use it as a tool if I'm designing something fairly complex.

Another tip is to remember that you can only divide any given piece of pie into equal sized segments at any given time. Just keep cutting those pie pieces in half until you get the subnet size you're looking for.

It's also helpful to write out the address ranges next to each piece of pie so you know where the stop and start. That will help to make sure you don't accidentally use overlapping ranges.

 

[rclough]

Thanks everyone for the advice, it's very helpful. Subnetting isn't one of my strong points and I don't want the class to pick up on it. I'll watch the video and work with pie example it makes sense with my thinking. If I'm right you do the same thing with the VLSM that you did with the subnet, if I'm wrong please tell me

 

[jneiberger]

VLSM is subnetting; subnetting is VLSM. Don't get too caught up on the terms. At the heart of it, all you're doing is dividing address space up into smaller, more manageable chunks. That's it. Or, conversely, perhaps you're combining to smaller address space chunks into a larger chunk. Either way, it's all VLSM/subnetting.

In fact, I think that the very term VLSM makes this sound much more complicated than it is. You're just dividing up a pie and handing out the pieces to the people who need it: some people need a big piece, some people need a small piece. Subnetting is nothing more than figuring out who needs which piece of the pie.

 

[vipergg]

Its become very obvious to me working in this field there are a lot of people who dont have a clue on this subject too . we recently took over an acount and everywhere they had subnets setup using a /16 boundary . They certainly had no clue about broadcast domains and how many users you should have on a subnet . In most cases a /24 or even smaller depending on forcasted users is a good size , certainly would not go bigger than a /23 .

 

[jneiberger]

That's true that you probably will never need or want a single subnet larger than a /23 on a LAN, but you will work with much larger blocks when dealing with Classless Interdomain Routing (CIDR). There's another acronym and term you can ignore, though: It's ALL subnetting whether you're dealing with subnet masks or prefix lengths. It's the same concept. I think people get confused because we call this stuff by names that sound really hard!

 

[burtsbees]

Especially when the explanation of CIDR shows the entire blocks of class A, B and C each as summarized, like class B with a /12 mask, for example.

Burt

 

[jneiberger]

That brings up another point: classful vs. classless thinking. Classful thinking and teaching should be abolished. It simply confuses the issue because people have to unlearn it when they get into the real world. There are no such distinctions any longer. Everything is referenced by prefix length, not class, and teaching new students to think classfully is pointless and counterproductive. It teaches them nothing that they need to know and just makes it harder to unlearn classful thinking in order to get on with classless thinking.

So, my advice to you when you teach your class: don't even bother teaching the classful approach. It has no benefit other than from a historical perspective.

 

[plshlpme]

i am starting to look at the bsci now for the ccnp.. and it doesn't mention IGRP or RIP. so even cisco is starting to only think classlessly...

 

[jneiberger]

Well, they're getting rid of outdated routing protocols, but I believe they still teach classful addressing, don't they? I'd be extremely happy if they removed classful addressing from their curriculum.

 

[rclough]

The new curriculum with Cisco Exploration mentions classful as "in the early days of IPv4". This chapter I'm about to cover with subnetting talks more about classless. I agree with getting rid of confusion when it comes to Cisco because it can be overwhelming with out teaching things to remember and then have to forget. What makes it really hard is I have a class full of future networking professionals who have every excuse in the book not to do the work and whine about failing.

I also have a book I'm reading through for the CCNA 640 - 801 exam that covers classful almost as much as classless. There's been so many changes it's amazing. Exploration for CCNA1 has the students working with routers right away and in the end of the course they work with media. Were working on things now that I did in semester 4 in 2005.

 

[burtsbees]

I too was in semester 4 in 2005...so now in semester 1 they study WAN technologies? Also, are you teaching anything about MPLS in CCNA? I continued my studies and went on to CCNP, and got it last December, right before they changed all the testing...I'm probably going to start studying soon to take one of the newer CCNP tests that covers VoIP that was NOT covered when I became CCNP. It's like the people that become CCNA's after taking your class will be equivalent to what CCNP is today!

Burt

 

[rclough]

Hey Burt...a lot with Cisco has changed since 2005. With the way the new material is I really think more high schools need to implement Cisco. From what I can tell the Discovery covers some of what we had. It's funny because I hated the class when I took it and now I'm teaching it.

So far I'm on chapter 6 and we haven't covered anything on MPLS. I've been thinking about going for my CCNA (I have to to keep my teaching certification) and from what I know I have about a year yet to test with the old curriculum. I'll be happy to get that far

 

[burtsbees]

Well, you could go for CCENT, which is the first two semesters of the old cirriculum, I think...

Burt

 

[uzumaki]

Try looking for Learkey CCNP-BSCI the old version on this forum. Mic Storm explains VLSM with ease. I too will be teaching soon this will be a good reference for the student. Goodluck.