The recommended settings for giga ethernet connected to Ethernet port

Question

What is the recommended duplex and speed configuration for giga ethernet port connected to legacy ethernet (10 mbps) port and there is no available duplex command available on the ethernet side? If we keep the giga ethernet side with auto it will negotiate to 10 Mbps half duplex and if we configure it with 10 Mbps full duplex we will get alarm about misconfiguration on the switches, so what is recommended?

Answer 1

The very short answer: don't configure anything.

Auto negotiation (or the lack thereof as Ron's detailed) works only when it's left alone. Manual settings can very easily cause problems either right away when done incorrectly or later on when hardware is upgraded.

For 1000BASE-T, auto negotiation is required - most hardware won't allow you to manually configure 1 Gbit/s for that reason. 1000BASE-T requires a clock master on a link that has to be (auto) negotiated. When a port can be configured to 1000BASE-T only, it's not a "use 1000BASE-T full-duplex no matter what" but an "auto negotiate with 1000BASE-T full-duplex as only option".

Most basic 1000BASE-T ports will auto negotiate down to 10BASE-T half or full duplex just fine. Some ports (SFPs or 10GBASE-T ports especially) don't support 10BASE-T any more however.

Usually, there will be no visible error message when the link is configured incorrectly. Speed mismatches simply don't link at all. Duplex mismatches are nasty - the link comes up and appears to work but it only does so at an extremely slow effective rate. There are error counters on the NIC and in managed switches to indicate the problem but it's not obvious.

So, don't configure manually. If you've got ancient equipment causing faulty auto negotiation (early Cisco hardware in particular), replace it.

Answer 2

In reality, the legacy 10 Mbps ethernet interface probably can't negotiate, and it can probably only do half duplex (very few 10 Mbps interfaces can do full duplex). You should let the 1 Gbps interface auto-negotiate. It will try to negotiate, but if the 10 Mbps can't negotiate, it will detect (not negotiate) that the connection is 10 Mbps, and it will set itself to half duplex, which is the default for 10 and 100 Mbps ethernet.

Cisco has a document with a table which shows what happens when different speed/duplex setting are configured on each end of a link: Troubleshooting Cisco Catalyst Switches to NIC Compatibility Issues

Why Is It That the Speed and Duplex Cannot Be Hardcoded on Only One Link Partner?

As indicated in Table 1, a manual setup of the speed and duplex for full-duplex on one link partner results in a duplex mismatch. This happens when you disable autonegotiation on one link partner while the other link partner defaults to a half-duplex configuration. A duplex mismatch results in slow performance, intermittent connectivity, data link errors, and other issues. If the intent is not to use autonegotiation, both link partners must be manually configured for speed and duplex for full-duplex settings.

Answer 3

A) offer only a subset of speeds

In cases where speed/duplex negotiation becomes difficult, you may want to consider the following to restrict a switch to offer only a (sub)set of speeds on a given port, while maintaining support for half/full duplex negotiation.

Please note that this is not a cure-all recipe, it just helps to narrow the scope of things that might go wrong.

The example is for Cisco IOS switches. This is certainly supported on the access switch types, such as 2960/3560/3750 series, and the younger 3650/3850 as well.

Other vendors may have similar features.

interface GigabitEthernet0/2
 ...
 speed auto 10
 ...

interface GigabitEthernet0/3
 ...
 speed auto 100
 ...

interface GigabitEthernet0/4
 ...
 speed auto 100 10
 ...

I remember running into issues of this type with crippled structured cabling where a 8-wire Cat5e was split into two 4-wire Cat5 links, combined with gigabit NIC and gigabit switchports.

The NIC driver of the OS was cabaple of helping the NIC to detect that only 2 wire pairs were available and to fall back to 100Mbps.

However, during PXE boot, no "proper" NIC driver was loaded, and the NIC still believed it had "line protocol up" at Gigabit - when actually, there were only 2pairs available on the wire. By consequence, DHCP and PXE would fail at this stage in the boot process. The cure was to configure the switch port to speed auto 100 10 (or to do the right thing and stop splitting an 8-wire cat5e into 2x4-wire).

B) Beware of flooded traffic

Another thing to consider when connecting 10Mbps devices to Gigabit ports and multi-gigabit backplanes: Beware of unknown unicast flooding and multicast.

When mixing 10M and 100M or even 1G devices in the same VLAN/Broadcast domain, be sure that it does not suffer from flooded traffic at elevated rates. An inadvertedly flooded multicast stream, or unknown unicast flooding (because of suboptimal L2 topologies) of anything >10Mbps will heavily oversubscribe the 10Mbit/s switchports, resulting in a DoS situation for the 10Mbit/s devices connected.

To prevent multicast flooding, look into IGMP snooping, to prevent unknown unicast flooding, check your vendor's LAN design guidelines; at the very limit, consider segregating the 10M devices into a broadcast domain/VLAN of their own.