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 have a 10/100/1000 24-port + 2 mini-GBIC ports switch which I have to link to other switches. Unfortunately, on the other switch all the mini-GBIC slots are taken up, so all I can do is link them together using a regular Ethernet port (which is also 1 Gbit/s) on the switch.

My question: does using a regular port instead of the uplink have disadvantages? If I understand correctly, the only advantage of that mini-GBIC adapter is that it is fiber so you can link switches that are kilometers away.

share|improve this question
3  
Care to share the make and model of the switch? –  ewwhite Jul 1 at 13:57

3 Answers 3

If your switch isn't non-blocking then often the uplink ports are non-blocking, so they could have that benefit, otherwise I can't think of any other real differences between types of port.

share|improve this answer

Some switches will have additional benefits when using the SFP uplink ports that are not readily apparent. For example, they could have larger hardware buffers dedicated to than available to normal ports. This is more important on uplink ports as they are more likely to see bursts of traffic that may exceed their capacity.

share|improve this answer
2  
Larger buffers is not always an advantage. Larger buffers means fewer lost packets, but it also means higher latency. In my limited experience, you are more likely to come across buffers which are too large than buffers which are too small. –  kasperd Jul 1 at 14:08
    
@kasperd, unless you are talking about a very time sensitive environment (for example, a trading floor), when you are talking about a LAN, larger buffers are an advantage to deal with bursts of L2 traffic. Or are you suggesting that a few micro-seconds are more damaging than a dropped frame (which with TCP traffic would then require a retransmission)? –  YLearn Jul 1 at 14:43
    
TCP expects to drop packets. Dropped packets are how TCP calibrates available bandwidth -- it increases the transmit rate until it observes dropped frames, then it backs off, then increases transmit rate until it sees dropped frames, etc. Generally on switches the "uplink" port is just another port that is identical to the others, even if it has an SFP physical link as well as a standard rj45 connector. –  chris Jul 1 at 14:48
    
@chris, you do realize that latency is also a factor in determining the TCP window, correct? So additional latency will also cause a "back off". A dropped TCP frame and retransmission has a number of disadvantages over slightly higher latency. For example, if the network is already congested, it adds additional frames on the network. It also requires that the packets be processed out of order and could delay the whole process until the retransmission is complete. –  YLearn Jul 1 at 14:53
    
@YLearn I wouldn't have noticed a delay of a few microseconds. It takes many hops with a few microseconds delay on each to add up to something noticeable. But 10s of milliseconds is something I see from time to time. It is mostly an issue with equipment designed to operate at a wide range of different speeds, they may pick a buffer size suitable for the highest speed and stick with that even if running at a lower speed. The most extreme case I have seen was a buffer large enough to hold one minute worth of data, but that wasn't on a switch. –  kasperd Jul 1 at 16:19

It should make next-to-no difference in performance. However, if the mini-GBICs are fiber ports and the other ports are copper links, you will lose galvanic isolation between the two switches. This can, in rare cases, cause a few more hardware failures.

share|improve this answer

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.