Damn Subnetting???/

[3steps]

No matter how many times I read about subnetting, I just don't get it. It could be that I space out midway through. Anyway, I was wondering if someone could help. Right now I have a network with 192.168.16.0 and 255.255.255.0 for ip/subnet. I would like to have 4 of the machines on the network, on their own subnet, but still be part of the large network. Can someone let me know how I would setup a router for this? and any router recommendations? Would a cheap cable router do?

Thanks
3Steps

 

[SaigonK]

You could do this: It would give you only four systems on that subnet.


Subnet Address : 192.168.16.0
Subnet Mask : 255.255.255.240
Subnet bit mask : 110nnnnn.nnnnnnnn.nnnnnnnn.sssshhhh
Subnet Bits : 28
Host Bits : 4
Possible Number of Subnets : 16
Hosts per Subnet : 14

 

[nightowl]

Actually, if you only need 4 hosts per network you can subnet it this way.

Initial network
192.168.16.0 / 24

Subnetted
mask: 255.255.255.248 or /29

networks
192.168.16.0
192.168.16.8
192.168.16.16
and so on

You can have 6 hosts on each network in this configuration. Also, what do you mean by "still be part of the large network"? You would need a router to do this. Some of the broadband routers support RIPv2 which you would need to transmits the network masks but I am not sure how well it would work.

 

[spidey07]

I guess the real questions is "why?"

For a home network everything can and should be on one subnet. Trying to do routing and whatnot only adds complexity that isn't really needed.

If all hosts are on one layer2 domain (switches) then they have to be on the same subnet with the same mask - 192.168.0.0/255.255.255.0 is just fine.

Can you explain what your trying to do?

 

[3steps]

This isn't for a home network. There are three machines in the network which access another machine very frequently. I was hoping by putting them on their own subnet, away from the 15 or so other machines, it could improve performance in a network run program. So basically, there would only be two network segments; one with these four machine, and one with about fifteen machines. I still need them on the same network, just different network segments. I hope this clears things up a bit. And thanks for all the help so far.

3Steps

 

[djNickb]

Just put them on their own switch if they're in close proximity to each other.

 

[spidey07]

actually putting them on different subnets (also called a network) would actually hinder performance because you'd have a router in involved.

Probably best to just throw all the machines on one switch.

 

[AFB]

Originally posted by: spidey07
actually putting them on different subnets (also called a network) would actually hinder performance because you'd have a router in involved.

Probably best to just throw all the machines on one switch.

<<--Agrees

Switch > router for speed if they are in the same class.

 

[xyyz]

speaking of switches and routers and segments.

can someone run through the different kinsd of segmentation you get with routers and switches.

routers provent broadcasts, while switches don't (looking at basic switches). what exactly do switches do again? i know the direct traffic only where it's supposed to go based on MAC addressing, but they propogate broadcasts... so what type of segmentation do switches allow for?

 

[narzy]

the easiest way I've found to understand subnets is to first understand IP address classes then tackle subnets. if you don't understand network numbering and the binary behind it your not going to understand subnets. so here comes narzy's crash corse in IP class
IP addresses are broken into four 8 bit segments making it a 32 bit addressing scheme. tho we read 128.0.0.0 as an IP address (this is not a real one just easy for demo) the computer reads it as 10000000.00000000.00000000.00000000 simple enough?

breaking in to 8 bit binary
8 bit binary gives you number assignments as fallows from bit 8 to bit 1
128 | 64| 32 | 16 | 8 | 4 | 2 | 1 | so depending on where the number fits in the byte determains its value if we use our previous example the 1 in the first octet is equal to 128.

how the HELL does this relate?
easy, it makes it real easy to explain IP classes
class A address octets begin with 0 ex they go from
00000000.00000000.00000000.00000000
to
01111111.11111111.11111111.11111111
translated
0.0.0.0
to
127.255.255.255
but in actuality its 126.255.255.255 cause the 127 range is reserved for loopback and diags.

next catch is availible networks with class A its n.h.h.h


class B addresses start with 10 ex
10000000.00000000.00000000.00000000
to
10111111.11111111.11111111.11111111

translated
128.0.0.0
to
191.255.255.255

class B is n.n.h.h

class C addresses start with 110 ex
11000000.00000000.00000000.00000000
to
11011111.11111111.11111111.11111111

translated
192.0.0.0
to
223.255.255.255

class C is n.n.n.h

and the last mention is Class D wich you probably guessed starts with 1110 ex
11100000.00000000.00000000.00000000
to
11101111.11111111.11111111.11111111

translated
224.0.0.0
to
239.255.255.255

class D is for multicast and is not used for networking devices generally.
h.h.h.h

make sense?

with subnetting your creating more network bits then designed into a classful system as above. but you have to fallow the same bit rules.

if I have a class B IP I by default have 1 network with 65,535 hosts

with subnetting I can break my network down into more manageable peaces by "barrowing" host bits and "making" them network bits say I want 255 networks with 255 hosts per network
with class B by default the bits look like this when comparing networks to hosts.

nnnnnnnn.nnnnnnnn.hhhhhhhh.hhhhhhhh

I want more networks, less hosts per network. so I set a subnet of 255.255.255.0 on my IP my network to host comparison then changes to look like this.

nnnnnnnn.nnnnnnnn.nnnnnnnn.hhhhhhhh

see what I did, I barrowed all the bits out of the 3rd octet and made them network bits.

now I can do this in any number of ways but I HAVE to barrow from left to right I cant have a subnet mask that in binary looks like this.
11111111.11111111.101101111.00000000

this limits what subnets I can have easiest way to figure it out is to remember the binary table up above

however if I want more networks I can "barrow" network bits
128 | 64| 32 | 16 | 8 | 4 | 2 | 1 |

 

[3steps]

Thanks for all of the help. I think I will go with a switch, since most agree that's all I need.
Narzy thanks for the crash course I'm going to save all of that, and keep going over it till I finally understand it! I didn't have time t read it all, as I'm at work, but hopefully that will help me with binary/ip addressing/and subnetting.

Thanks again,
3Steps

 

[ITJunkie]

Originally posted by: xyyz
speaking of switches and routers and segments.

can someone run through the different kinsd of segmentation you get with routers and switches.

routers provent broadcasts, while switches don't (looking at basic switches). what exactly do switches do again? i know the direct traffic only where it's supposed to go based on MAC addressing, but they propogate broadcasts... so what type of segmentation do switches allow for?

A switch gives you "wire speed" connections. They tend to move packets faster in other words. If you throw in a layer 3 capable switch you can add some routing capabilities, ie IP based Vlans (network segments).

 

[spidey07]

segment = collision domain (switch breaks up the collision domain so a 12 port switch has 12 segments)
network = broadcast domain (router breaks up broadcast domain)

hub = layer1
switch = layer2
router = layer3

hope that helps.