Subnetting a network

oddjob62 · 16 Sep 2008 at 22:23

Quick question in the hope that someone's done this before and can let me know if there's any "issues" i may run into.

Currently our head office network is one big class b network. Now as part of the office move, we are getting a completely new network infrastructure as well as a voip phone system installed. I decided this would be a good time to partition out our network, separate voice, client, server vlans for example.

Now I though i could just create smaller subnets of our class b network (and have vlans set up accordingly), and have it separated so that our servers will keep the ip addresses they currently have (avoiding any issues with things mapped by ip instead of hostname for example). Keeping all IPs within our existing class b network would have the added bonus of not having to alter the routing tables on our WAN (outsourced MPLS WAN with 10 other offices). I would expect that all that would need to be done is to change the subnet mask and gateway on the servers and assuming routes have been set up correctly on the switches everything should carry on as normal...

Or is there a glaringly obvious flaw to my plan?

oddjob62 · 16 Sep 2008 at 23:09

Cheers for the speedy replies guys

Chief Wiggum said:
yep - that's about the gist of it, you have to give it some considerations though, for example size for growth and plan for supernetting, so that you can consolidate groups of your smaller networks behind a larger mask.

Yes, I've allowed for at least 300% (complete overkill) increase on all new subnets.

( |-| |2 ][ $;12519360 said:
Make sure none of the currently assigned IP's are assigned to the network or broadcast ip of the new subnets. I guess that's not a problem if you use dhcp but with static servers it could be.

Good point, yes, none of the staticly assigned IPs are on possible nw or bc addresses (guy who originally designed the network at least knew enough so as to avoid .1 and .255 addresses). Clients and Phones will all be DHCP.

Any more for any more?

oddjob62 · 21 Sep 2008 at 21:29

Laser402 said:
Remember a .1 would be usable its .0 your thinking off.

Better to not use an ip that can be used than use an ip that can't, but yes thanks for the correction.

oddjob62 · 22 Sep 2008 at 14:22

IAmATeaf said:
Depends entirely on the netmask you use to sub-net the lan as to what addresses will be nw and bc addresses, won't it?

Yes, although as i'm not using anything smaller than /24

oddjob62 · 22 Sep 2008 at 18:56

I think IAmATeaf was pointing more to the fact that .0 isn't always the network, but you are right about .1 as far as i can see.

Anyway... stop this silly argument and figure out if i may have an issue :mad:

oddjob62 · 23 Sep 2008 at 17:59

R4z0r said:
Also, if you were just using a flat L2 network before, make sure none of your apps are relying on L2 broadcasts to function properly and if they are you will need to setup "ip helper" addresses (Assuming Cisco kit).

Yeah, good point, might wireshark the network for a few days to make sure, cheers. BTW won't be Cisco, will be Extreme (no idea what their equivalent of ip helper is, gonna start reading up on it soon)