Server Fault is a question and answer site for system and network administrators. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

What disadvantages have to set the MTU to the maximum possible value in ALL network adapters on a network (including virtual switches vmware esx)?

Why do not the producers as a default?

I do not understand the way to use a network card to 1500 when it already support the value of 9000!

share|improve this question
up vote 6 down vote accepted

This is not a default because you would need equipment at every step in the chain to also support jumbo frames (MTUs over 1500).

Not every NIC or network switch or router or firewall supports such large MTUs natively.

share|improve this answer
Ok, but MTU stands for Maximum Transmission Unit! The Transmission Unit is not agreed upon during the handshake? – CSG May 13 '13 at 14:44
Just curious, to which particular handshake are you referring? The Ethernet handshake? – MikeyB May 13 '13 at 15:13
I'm sorry for my english.. "handshake" is not the correct term.. as it is called the phase in which the rate is established, MDI / MDIX etc etc ... between devices? – CSG May 13 '13 at 15:23
@CSG MDI/MDIX is "detected", not agreed upon. MTU is fixed administratively, there's no agreement protocol. The OS has MTU back-off mechanisms, based heavily on ICMP (which can fail because lots of very dumb ISPs block/limit ICMP in ways they definitely should not). MSS is part of TCP and is agreed upon during the TCP Handshake, and can be modified later during the session, and can be different per direction of the connection as well. – Chris S May 13 '13 at 17:17

Others have answered most of this, but as for the default, it's not the default because it wouldn't be backwards-compatible and there's no way to do any form of auto-detection in an automatic way.

It's also not that useful for most workloads. Storage and similar bulk transfers are about the only thing that benefit, and these generally stay within a subnet or small section of a network where managing the configuration is not a huge overhead.

share|improve this answer

everything within a VLAN MUST have a matching MTU size. Your routers MUST also be capable of sending ICMP fragmentation needed errors (most do, automatically) so that hosts can discover path MTU and successfully exchange data to hosts that are not running jumbo frames (such as internet destinations).

This doesn't just go for jumbo frames, it applies to everything - someone on cable internet with an MTU of 1500 can talk to someone else on PPPoE ADSL with a 1492 byte MTU because pMTUd is what enables the cable internet user's host to know that they can't receive frames > 1492 bytes.

share|improve this answer
Of course, then there's pMTUd blackholes which don't do this, and we end up with the terrible TCPMSS hack – Olipro May 13 '13 at 12:54

Your Answer


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.