Server Fault is a question and answer site for system and network administrators. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

We discuss that topic but we have too different interpretations. That is why I'd like to know how to resolve this issue. What is the truth? Can I define max 64 subnets or max 4?

Writing down an opinion will not suffice, sorry. We already have 2. We need to scientifically show (for example with official network standards) which is true.

share|improve this question

marked as duplicate by Iain, sysadmin1138 Jul 13 '11 at 13:09

This question has been asked before and already has an answer. If those answers do not fully address your question, please ask a new question.

Sounds a bit like homework, no? – MDMarra Jul 13 '11 at 12:57
@MarkM, sounds more like two programmers who don't know IP. – Chris S Jul 13 '11 at 13:00
sounds like 2 programmers who know IP but remember differently because at least one didn't understand one small detail of a protocol. Common human situation. – erikb Jul 13 '11 at 13:21
"Closed as exact duplicate" where's the duplicate link? – tombull89 Jul 13 '11 at 13:36
I must have included it when I started the voting as you have to – Iain Jul 13 '11 at 14:07

A /26 network has 64 addresses. Since the definition of a subnet is that it contains at least 2 addresses, the technically correct answer is 32. However, for practical implementations the actual answer is more like 16, since in order to be really useful, 2 addresses in each subnet (the network address and the broadcast address) should not be used for other purposes, and therefore a useful subnet comprises 4 addresses.

share|improve this answer
this is one oppinion. Can u prove that with sources please? Other people think it is different. To name it they think the x/26 means that there are 2 bits specifically defined to be used for subnets, and thus there are only FOUR subnets possible with 64 adresses (or 62, doesn't matter for this discussion) each. – erikb Jul 13 '11 at 12:21
is this a homework question? – SpacemanSpiff Jul 13 '11 at 12:24
My understanding of a /26 network is that the 26 highest bits of any address pertaining to the subnet are fixed (as per the standard definition of classless networking, CIDR), which then leaves the remaining 6 bits for hosts or subnets within the network. – wolfgangsz Jul 13 '11 at 12:26
erikb, wolfgangsz' response is not an opinion, it's simple math. See massimo's reply for a textbook-quality explanation. – Jodie C Jul 13 '11 at 12:28
Please read for details. This article contains numerous links to the various RFC documents going to extreme detail. If you get taught anything different, then your teachers are out of sync with reality. – wolfgangsz Jul 13 '11 at 12:37

You can create:

  • 1 subnet containing 64 addresses (/26)
  • 2 subnets containing 32 addresses (/27)
  • 4 subnets containing 16 addresses (/28)
  • 8 subnets containing 8 addresses (/29)
  • 16 subnets containing 4 addresses (/30)
  • 32 subnets containing 2 addresses (/31)
  • 64 subnets containing 1 address (/32)

Given that a IP subnet always has two reserved addresses, the lowest one (network) and the highest one (broadcast), and so needs at least two more addresses to be actually usable, the lowest number of addresses you can have in a "real" subnet is 4.

Of course, you can also mix them, because every subnet can be re-subnetted; so you could f.e. have a /27 subnet using up the first 32 addresses, and two /28 subnets using the remaning 32 ones, dividing them in two blocks of 16 addresses. In any case, by using more smaller subnets, you are going to waste more addresses (due to each subnet requiring two reserved ones).

As usual, you can find everything you need to know (and much more) here.

share|improve this answer
I refer u to my comment on the other post. – erikb Jul 13 '11 at 12:26
See edit and link. This is classless IP addressing, so you can subnet any network as much as you like, and then it can be re-subnetted again, and again, and again, until you use up all addresses. – Massimo Jul 13 '11 at 12:29
what is specific about classless subnetting? maybe that is the problem in my group. – erikb Jul 13 '11 at 12:44
@Massimo, PtP links may use /31 on OSes that support it. See RFC 3021 for details. (Almost every major OS and networking device supports /31 except Windows) – Chris S Jul 13 '11 at 13:07
@Chris, I know this but I preferred not to introduce too many details and edge cases in my answer ;-) Thanks for pointing it out anyway :-) – Massimo Jul 13 '11 at 14:37

Given a /24 subnet mask and allocating an additional 2 bits for subnetting to create a /26 subnet mask means that you've allocated 2 bits (from the host portion) from which to create 4 subnets of the original /24 network. But, that doesn't mean that you can't subnet the /26 subnet further... you can. It seems that you're assumption (or proposition) is that you can't subnet any further, and for the purpose of the debate you're trying to settle that may be true if the debate asks the question "How many subnets can I create with a /26 subnet mask", without considering the possibility of further subnetting.

The /24 is the original network, it is not a subnet (although it could be a subnet of a larger network). The /26 mask allows you to create 4 subnets from the original /24 network.

share|improve this answer
The last part is what my classmates say. Btw the maybe weird thing is that I actually have the oppinion that u can create 32 subnets (minus the bit for network id and broadcast id). I want to prove successfully that it is not my stupid oppinion (I'm the only one who thinks that) but the truth. So, ur last sentence actually sounds like what the other people say. Can u explain further why u say 4 now and not 32? – erikb Jul 13 '11 at 13:05

Not the answer you're looking for? Browse other questions tagged or ask your own question.