High broadcasts...check out our addressing scheme

Status
This thread has been Locked and is not open to further replies. Please start a New Thread if you're having a similar issue. View our Welcome Guide to learn how to use this site.

99ls1z

Thread Starter
Joined
Jan 11, 2006
Messages
3
I calculated the broadcast percentage on all of switches in our network. We have about 30 3Com switches. According to my caculations most of them are averaging 50% - 80% broadcast traffic except for our 2 layer 3 switches with have about 2% broadcast. I took the total amount of packets on each port and the total amount of broadcast on each port and divided giving me this percentage. All of our switches are in the ip range of 10.10.33.0 255.255.255.0. We are pushing 10.10.34.0 10.10.35.0 10.10.36.0 through DHCP to workstation, printers, and statics for servers. We are using a total of 4 subnets and they are routed through 2 layer 3 switches we have that are on opposite ends of our campus. So pretty much on every switch that we have you have workstations connected to it that on are each of the subnets depending on which address they get through the DHCP lease. Our servers have a static address assigned from each of the subnets. It seems to me that on each switch it is getting broadcasts from 4 different subnets. I am just seeing a high amount of broadcast traffic on our network.

This is a suggestion that I have. Take 1 subnet and use for our switches and servers. We have 2 layer 3 switches that are on opposite ends of our campus that basically splits it in half. Take 1 subnet and put it behind the layer 3 switch on the back of campus. Take 1 subnet and put it behind the layer 3 switch on front of campus. This way every device that is connected to a switch is on the same subnet and the layer 3 switch will provide the routing between the subnets allowing the workstations to access the servers which are on a different subnet. I just thought this makes more sense than having for example 3 workstations connected to the same switch but on different subnets. Could this be causing all the broadcast traffic?
 
Joined
Aug 26, 2005
Messages
894
That's not uncommon. But, by for your broadcast count, what's the actual payload and datarate? Windows has a bad habit of poomping NETBIOS NS requests, whois, etc into the subnet that it is attached to - so while it may be alot of broadcasts, the payload size of each broadcast may be small. Also, considering that you have 30 switches I would say the majority of these broadcase frames are actually IGMP or BDPU messages from the switches themselves (for spanning tree) Please tell me spanning tree is running... :)

You do have a good idea about splitting your networks from each respective L3 switch. I'm guessing you have them paired together and are running OSPF or somesuch, so that'll give you some redundancy in the event that a L3 switch fails.

One thing I wouldn't suggest is attaching devices to the MGMT network of your switches. One of the reasons you do want to limit this IS broadcast traffic. CPU's on switches aren't as fast as a PC's CPU, and performance can be affected if the switch has to process and discard broadcast traffic that happens to be on it's subnet.

Another thing to consider would be your uplinks. what kind of ports are they? What's your network topolgy? For this campus, does each switch connect to a headend L3 switch, or do you have some sort of ring/daisy chain happening?
 

99ls1z

Thread Starter
Joined
Jan 11, 2006
Messages
3
We do have spanning tree running. All of our switches are 33.0 subnet. All of our servers have a static 33.x address and we clients getting 33.x from DHCP. Our servers have a static ip assigned from each of our subnets. I ran ethereal on my computer today. I have a static ip assigned in the 34.0 range. Within 1 minute I got over 600 arp broadcasts packets. I was getting packets from servers looking for computers on 33.0, 34.0, 35.0 subnets. Our main L3 switch has an ip from each subnet assigned to its interface. The L3 switch is the default gateway. The default gateway on the L3 goes to our ISA server then PIX, Router. We basically have a front and back side our campus separated by a lake. The L3 I am talking about is located on the front of our campus. On the back side of campus is our other L3. There are two aggregated gig links connecting both ends.

In our building we have 3com 4400 SEs connected back to 4400 FX. The 4400 FX is connected to the L3. We have a 3Com 3824 connecting our servers which is a gig switch.
 
Joined
Aug 26, 2005
Messages
894
When you say your servers have statics on all three subnets, does that mean you have a server with a primary and two alt IP's?

(i.e. 10.10.33.10 / 24, 10.10.34.10 /24, 10.10.35.10 /24, etc) If this is true, are you running AD? How the heck is DNS working? Is AD only registering one IP?

Sounds like you're not doing much with the L3 switches. It's just your Internet gateway. Baah.

So, here's a cheezy text diagram. Forgive me if some IP's are incorrect for you network, I'm just guessing. - But, your thougths are pretty much correct. You just have to create a VLAN for each network, and the L3 switches have to talk to each VLAN. This can be done with physical port or .1q trunk.

<cheezy_diagram>
Internet (connect to PIX outside)
|
PIX inside 10.10.1.1 - OSPF 1 area 0, receive routes from L3SW1
|
| ISA server 10.10.1.2 - static route 10.0.0.0/8 via 10.10.1.3
_____|
| 10.10.1.3
L3 SW 1 <- OSPF 1 area 0 10.11.255.0 /30 -> L3 SW2 (GIGE)
4400FX switch connect to each L3 sw, allow VRRP to run. Create VLAN's with appropriate ports, or .1q to each L3 sw for appropriate VRRP network.
VRRP VLan 34 10.10.34.1 (2 and 3 used for VRRP L3SW1/L3SW2)
VRRP Vlan 35 10.10.35.1 ("")
VRRP Vlan 36 10.10.36.1 ("")
Connect distribution switches to to port in appropriate VLAN on 4400.
</cheezy_diagram>

------------------

I'm curious, are you running an NBX phone system? I can't think of any other reason to have 3COM switches.
 

99ls1z

Thread Starter
Joined
Jan 11, 2006
Messages
3
We don't have a NBX phone system. I think 3COM was purchased due to price issues. You are correct about the servers. They have a primary and 2 secondary ips. The tech that setup are network designed so that the front and back of our campus are connected with fiber. The L3 in the back and L3 in are connected. If this links breaks than only the front half of our campus could have network access. The gateway on the L3 in back of campus is pointing to the ip of the L3 in front of campus. There is no routing protocol being ran on the L3 switches.
 
Status
This thread has been Locked and is not open to further replies. Please start a New Thread if you're having a similar issue. View our Welcome Guide to learn how to use this site.

Users Who Are Viewing This Thread (Users: 0, Guests: 1)

As Seen On
As Seen On...

Welcome to Tech Support Guy!

Are you looking for the solution to your computer problem? Join our site today to ask your question. This site is completely free -- paid for by advertisers and donations.

If you're not already familiar with forums, watch our Welcome Guide to get started.

Join over 807,865 other people just like you!

Latest posts

Members online

Top