Question

  Asked: Apr 2 2007   2:26 PM GMT
  Asked by: kevins74


Looking for clarification on new IP layout


Career development, Networking, Availability, Hardware, Routers, Switches, Hubs, Cabling, Cisco, Dell, Protocol analysis, Network monitoring, Network protocols, Ethernet, IPv4, TCP, Networking Services, DHCP, DNS, Active Directory, Microsoft Windows, Tech support

I currently have a 192.168.1.0/24 network. I have one cisco router with many cisco switches that allow everyone to have a DHCP ip of 192.168.1.X. I will soon have more people then the /24 will handle, so I would like to switch to:

172.16.30.0 subnet
255.255.254.0 mask
172.16.31.255 broadcast

That gives me as my starting IP, 172.16.30.1 and an ending IP of 172.16.31.254. Does this mean that i can have a swich with 172.16.30.4 and another switch with 172.16.31.19 and they are both on the same network and will be able to talk to each other just like if they were 192.168.1.10 and 192.168.1.11 ?

I am trying to keep the network as simple as I can. I am not interested in creating Vlans and stuff like that.

Also, with my current DHCP, I have the scope as start IP 192.168.1.1, End IP 192.168.1.254, Subnet mask 255.255.255.0 then the address pool has the range for distribution as the entire 1.1 - 1.254 with another range of 1.1- 1.99 being excluded from distribution.

Setting up the new DHCP should be about the same with the new IP's. Scope, range for distribution, range to be excluded, etc...

Any help is greatly appreciated.

Thank you

Subscribe to Alerts! Get questions and answers delivered to your Inbox.


E-mail me updates on this question



   SUBSCRIBE

hidden modal window

Answer Wiki (Improve, edit or add to this answer)


 RATE THIS ANSWER
0
Click to Vote:
  •   0
  •  0



Yes, the two IP's 72.16.30.4 and 172.16.31.19 would be in the same subnet. You could also setup 192.168.x.x with a 255.255.255.254.0 address and achieve the same result, as the 192.168.x.x range is a 16 bit subnet mask. This may be less headache since you're used to the address space already. Change all of your subnet masks to reflect the new one and add new hosts in the second address range you choose to migrate the network.

Just realize that the reason to avoid making such a large subnet is because of broadcast storms that may be experienced with that many hosts in the same collision domain. Every system that receives a broadcast message must process the message, regardless of whether it is addressed for that machine or not. That is the reason that most people subnet their networks, not because they want to make it more complex. IPv6 is supposed to fix the broadcast storm problem by using multicast addresses versus using ARP to announce layer 2 addresses and find other computers.

Don
  • AddThis Social Bookmark Button

Browse more Questions and Answers on Networking, DataCenter and Storage.

Looking for relevant Networking Whitepapers? Visit the SearchNetworking.com Research Library.


Discuss This Answer


You must be logged-in to discuss a question. Log-in/Register

kevins74  |   Apr 3 2007  9:39AM GMT

Don,

Thanks for the response. With that said, do you think I would really see that much of an impact with using the 172.16.30.0 subnet with a 255.255.254.0 mask. I do not plan on this getting any bigger at least with this office and I can’t even see it using much of the extra range. All my switches are 3560’s and I am fine with keeping the subnets small, it is just not at this moment in time (a lot going on)

 

Swiftd  |   Apr 3 2007  8:45PM GMT

I’m sure you’re fine. I’ve heard the theoretical maximum number of hosts per an ethernet network is 100 hosts. Of course, this varies widely with how chatty your hosts are. Check out this white paper. I’ve personally done more than 100 hosts per network, as many people have. I’m just saying that’s what the paper says.

<a href="http://www-lt.fe.uni-lj.si/vaje/viri/seminarji/vaje_2002-2003/otk2/clanki_pdf/Measured%20Capacity%20of%20an%20Ethernet%20-%20Myths%20and%20Reality.pdf" rel="nofollow">http://www-lt.fe.uni-lj.si/vaje/viri/seminarji/vaje_2002-2003/otk2/clanki_pdf/Measured%20Capacity%20of%20an%20Ethernet%20-%20Myths%20and%20Reality.pdf</a>

I wonder how old and accurate the paper is. Nonetheless, Ethernet hasn’t changed drastically over the years. Only the cabling has changed.

You’d have to test it in your environment to see if you get bottlenecks and broadcast storms when you put the two segments together.

HTH,
Don

 

astronomer  |   Apr 10 2007  2:04PM GMT

Kevin;
I have a similar issue as we restructure one of our campuses. In a modern network the limit is mainly determined by broadcast traffic and this is highly dependent on the operating systems used. I put the question to HP procurve support and cisco, “How many nodes should I limit each subnet to?”

Cisco didn’t respond but HP told me a modern switched net like ours, (1 Gig backbone, 100 Mbit to clients, windows 2000 and XP), shouldn’t have more than 300 nodes on a specific subnet. On our main campus our /22 nets are suffering from excessive broadcasts. Your planned /23 net is half as big as the ones I am trying to get rid of.

Given this response, I will be limiting each subnet to class C. I suggest you follow a similar model and use two class C subnets connected by a router or layer 3 switch. You can avoid VLANs if you wish by partitioning physically with each switch on a specific subnet. This a more complex design than you have contemplated but it will allow cleaner growth. If you do pull the mask out to supernet two class C nets together, keep a watch on your broadcast traffic. As other responses suggested, there is no reason you need to switch to the 172.16 range with modern operating systems.
rt