Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

I'm having two ip networks (192.168.1.x and 192.168.2.x) bridged on a server, but all hosts (fixed IPs) are on the same physical 100BaseTX ethernet (with a daisy chain of 48ports switchs).

Often, I loose link between the server and hosts of the second network. Usually if I reboot the host, the connection work again, and if I force connection to be active, the connection keep alive until I ping after a while without active connection.

I'm wondering how many hosts can be connected on the same network without troubles, and eventually how many switches can be daisy chained ?

Any comments on how to find a solution to the problem and not answering directly to the question are welcomed!

share|improve this question
    
As well said by joeqwerty, the question may not resolve the problem. This thread expects an answer to the question and not to my specific problem ! –  snowflake Jun 18 '10 at 15:55

3 Answers 3

If possible you shouldn't daisy chain switches, but arrange them in a tree. Get a good primary switch and connect the server, the most important devices, and the other switches to it. Then use the other switches to serve the bulk of your network.

This will also ease cabling allowing you to run a single ethernet link to a 'secondary' switch near a group of devices (eg, one switch per floor / department / room are common realities).

When LANs grow large the differences between a quality switch and a cheap one will start to emerge.

Also don't use software bridging unless you really need to. Are you that short on switch ports / ethernet cables?!? Other than performance issues, a pc is much more prone to failure than an ethernet switch.

share|improve this answer

Your question is a good one from a theoretical point of view, but not likely to elicit a solution for the root cause of your problem. You're using the server as a network bridge, when the problem occurs a reboot of the server resolves it. My conclusion is that the server (acting as a bridge) is the real cause of the problem and is where you should focus your efforts on finding a solution. Do you have a router or firewall (which will route traffic between interfaces) that you can put in place of the server?

share|improve this answer
    
Reboot of the host resolves the problem ! not reboot of the server. Using a router is an option for diagnostic (and not sure for final solution due to the cost, since it is a kind of architecture described here that is multiplied in multiple instances) –  snowflake Jun 18 '10 at 15:54

Number of 'hosts' (really a term for hypervisor these days, I suspect you just mean 'devices') is usually very high but in real terms it's only limited by your switch CAM table size. You'd right about the 100m single link issue. As for limits to 'daisy-chaining', well although I don't endorse it the rules for hubs and switches are different - can you give us more/better information as this is quite a hard question to read/understand.

share|improve this answer
2  
Hosts is still widely used in the traditional sense as well, especially in documentation. –  MDMarra Jun 16 '10 at 11:07

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.