DHCP and IP Addressing

[NEM1S1S]

I am currently in the process of rebuilding my LAN (300 PC's, Windows 2000 SP4 connecting to Novell Netware 6 servers). I have currently implemented Class A addressing (10.x.x.x addresses). I don't know alot about the difference between Class A, Class B, Class C. Could someone enlighten me on what is better and for what reasons.

Also, does anyone in here know how to configure DHCP under Netware so I can have multiple ranges...thanks

[TangZ]

I dont know about the ranges but as for the different classes (a,b and c) an a class ip means that the first octet is the network and the other three octets are the host. class b - first 2 octets network, last 2 octects host, c - first 3 octet's network, last octet host.

there is really no difference in the addressing schemes besides the fact that most class a addresses can be subnetted into alot more ip's then the other 2 classes.

most dialup connections use the class c address whereas broadband uses class a.

I honestly dont remember ever seeing any ip I have picked up dynamically being a class b..? thats a strange thought..

where DO you see class b addresses...anyone got an answer to THAT one?

[BigFred]

Hi NEM1S1S

If you are using private addresses on the inside, it doesn't really matter what class you use, provided there are enough available IP addresses for the number of machines in your network.

What you've got is fine. And by using class A you have the potential to split your internal network into a number of smaller sub-nets if you so desire at a later date, as you have lots of addresses to play with.

TangZ, some large organisations use class B addressing. My employer does this. The network is subnetted into a number of subnets, each being a class C network, in it's own right. This is achieved by each subnet having a unique IP number in the third octet. Each subnet uses a DHCP server of it's own. I am administrator for one of these subnets. Private IP's are not used. BF

[NEM1S1S]

BigFred, so basically it dosen't really matter weather it's Class A, B or C? cool.

[mankyway]

Class A addresses range from 0.0.0.0 to 127.0.0.0
Class B addresses range from 128.0.0.0 to 191.0.0.0
Class C addresses range from 192.0.0.0.to 223.0.0.0

Classes D & E use 224.0.0.0 & 225.0.0.0 respectively for multicasting and scientific purposes.

With the aforementioned 3 classes, you subnet to your network requirements.

Why have you chosen 10.0.0.0? You canuse this but usually private networks utilize 192.168.0.0 and have Network Address Translation for any internet traffic.

I am not sure about the Novell DHCP Server, but most servers, you specify a starting IP Address and then how many addresses you want in that DHCP Pool.

Tangz, that part about the dial up and Broadband is misleading. Each ISP has a block of IP Addresses assigned to them and they subnet them to use both Dial-up and Broadband.

B class addresses are used by some ISP's but it is a case of getting as many host addresses out of your subnets as possible as IP addresses equal money!

With the original question, Nem1s1s, have you planned your new network, with regards to subnetting etc? ie How many Routers do you have?

[Tuttle]

Class A/B/C is fairly archaic terminology these days anyway, very few public networks are broken up on those boundaries anymore. They're certainly not used when allocating network ranges to ISPs.

Once upon a time the classes had very specific meanings - as well as the address ranges mankyway posted, class A addresses were always used with a 255.0.0.0 network mask, class B with 255.255.0.0 and class C with 255.255.255.0. Then the addresses started running out and subnetting was introduced, followed by variable-length subnet masks. CIDR (classless inter-domain routing, where blocks could be allocated by registries with arbitrary subnet masks) followed later, and pretty much abolished any significance which class A/B/C had left.

10.0.0.0/8 is actually a pretty good choice for larger private networks IMHO - lots of room to subnet if you have different sites, and plenty of room for those administrative niceties, eg:
10.x.y.z/255.255.0.0
x represents the site
y represents the type of equipment (1 = infrastructure, 2 = server, 3 = DHCP pool etc)
Highly inefficient, but it's a private network so you can do what you like.

If you want to read up on the finer points of IP addressing, subnetting etc, http://www.3com.com/other/pdfs/infra..._US/501302.pdf is a good read.

[NEM1S1S]

Ok, what happened is that I had my subnet mask wrong. I had it as 255.255.255.0, I re-created my DHCP as a 10.x.x.x subnet address and a subnet mask of 255.255.252.0 and then it gave me three ranges to play with. I excluded the first range for servers, printers, switches (I will hard-wire these)

[shalomb18]

Class A allows 2^24(2 to the power 24) hosts per network because the default mask is 255.0.0.0.
Class B allows 2^16 hosts per network (255.255.0.0).
Class A allows 2^8 hosts per network (255.255.255.0).

Now the subnet mask helps any device running IP to identify the network/subnet portion of an IP address, so if you had you network running on the range 10.16.2.x /22 (SM-255.255.252.0), the devices know that they belong to only the 10.16.0.0 network.....If you dont understand this, refer to a few websites that explain subnetting.

It doesnt actually matter what Class of addresses (private) you use as long as you have subnetted the Range to meet your requirements, e.g, I could use the Network 192.168.0.0 but use a subnet mask whose value is less than the default subnet mask .e.g 255.224.0.0....It does matter what class you use, if you use a class A network say 10.0.0.0(255.0.0.0), then you are allowing 65,534 hosts on that subnet but it is a physical impossibility to implement this coz you you will have bandwidth problems and a never ending "packet collision storm".. That is why subnetting was invented to use smaller networks...

Now what i dont understand is "I had my subnet mask wrong. I had it as 255.255.255.0, I re-created my DHCP as a 10.x.x.x subnet address and a subnet mask of 255.255.252.0 and then it gave me three ranges to play with...". You have created the "three" that you need Subnets, true, but also given yourself more hosts per subnet (1022 to be precise). You could use a subnet mask like 255.255.255.224 and have 8 subnets (Usually 6 on old networking equipment) with 30 hosts per subnet.

Mankyway, I have to correct you regarding Classes D and E, the ranges for Class D and E are 224.0.0.0 to 239.255.255.255 and 240.0.0.0 to 255.255.255.255 respectively.

[shalomb18]

On a Class C network this would be the data needed to subnet it.
The Last octect is the last part of the subnet mask you will use, these are the only possible values.

Last Octect No. of subnets No of hosts Prefix notation
.128-------------2* ------------ 126**-----------/25
.192-------------4*-------------- 62**-------------/26
.224-------------8*-------------- 30**-------------/27
.240------------16*--------------14**-------------/28
.248------------32*---------------6**--------------/29
.252------------64*---------------2**--------------/30
.254-----------128*--------------0 ***------------/31
.255-----------255*--------------1**** -----------/32

* - The formula to calculate the Number of possible subnets is actually (2^N)-2 on equipment running Older IOSes. N is the number of bits you use when encroaching the host part of an address to extend the no. of network bits, which is subnetting. If you use this subnet mask 255.255.255.128, you have 'borrowed' 1 bit from the host portion and theoretically that would be -1 possible subnets because the ZERO-SUBNET and BROADCAST SUBNET are supposedly unusable and tricky to use but Newer IOSes and OSes support the use of these subnets, so we conclude that 2N subnets are practically possible nowadays.

** - The number of hosts is (2^n)-2, n being the number of remaining host bits after some having being borrowed for 'Subnetting'. The host with all zero bits is the Subnet no. for that subnet and host ID with all 1's is the Broadcast address for that subnet, so they are unusable.

*** - No of remaining host bits, n = 1 , therfore 2^1-2 is Zero, also because the only possible 'hosts' on this subnet are infact a Subnet No. and BroadCast Address, unusable.

**** - The 2^n-2 rulse doesn't apply here, it is possible to use this as a valid host denoting that the hosts that recieve this Subnet Mask are treated as individual networks, The default gateway is always the same as the IP address, these are typically used over WAN PPP or Serial Links where a lower network layer Link/Virtual-Circuit is needed to transport data to/from this address eg PPPoATM. Most ISP's use this method nowadays to conserve address space or given it a 'VLSM' mask if the address was not being used on an already subnetted netowrk.