TCP segmentation offload

Linux specific questions/information are gathered here. The main thrust of topics are applied to Centos/RedHat(RH)/Debian/Ubuntu/Gentoo distributives

TCP segmentation offload

Postby lik » Sun Jul 18, 2010 5:38 pm

Abbreviated as TSO, TCP segmentation offload is used to reduce the CPU overhead of TCP/IP on fast networks. TSO breaks down large groups of data sent over a network into smaller segments that pass through all the network elements between the source and destination. This type of offload relies on the network interface controller (NIC) to segment the data and then add the TCP, IP and data link layer protocol headers to each segment. The NIC must support TSO.

TCP Segmentation Offload (or TCP Large Send) is when buffer's much larger than the supported maximum transmission unit (MTU) of a given medium are passed through the bus to the network interface card. The work of dividing the much larger packets into smaller packets is thus offloaded to the NIC. More specifically, the e1000 driver is passing 64k packets to the network card, which then divides these into proper MTU-sized 1500 byte packets.

To disable TSO for a specific NIC:
Code: Select all
ethtool -K eth0 tso off

And check the result:
Code: Select all
ethtool -k eth0

Offload parameters for eth0:
rx-checksumming: on
tx-checksumming: on
scatter-gather: on
tcp segmentation offload: off
lik
Founder
Founder
 
Posts: 497
Joined: Wed Dec 15, 2010 3:21 am

Setting NIC's speed parameters with ethtool

Postby lik » Wed Sep 01, 2010 3:45 pm

Setting NIC's Speed Parameters with ethtool

Unlike mii-tool, ethtool settings can be permanently set as part of the interface's configuration script with the ETHTOOL_OPTS variable. In our next example, the settings will be set to 100 Mbps, full duplex with no chance for auto-negotiation on the next reboot:
#
# File: /etc/sysconfig/network-scripts/ifcfg-eth0
#
DEVICE=eth0
IPADDR=192.168.1.100
NETMASK=255.255.255.0
BOOTPROTO=static
ONBOOT=yes
ETHTOOL_OPTS="speed 100 duplex full autoneg off"

You can test the application of these parameters by shutting down the interface and activating it again with the ifup and ifdown commands. These settings can also be changed from the command line using the -s switch followed by the interface name and its desired configuration parameters.
Code: Select all
ethtool -s eth0 speed 100 duplex full autoneg off

The Linux man pages give more details on other ethtool options, but you can get a quick guide by just entering the ethtool command alone, which provides a quicker summary.
# ethtool
...
...
ethtool -s DEVNAME \
[ speed 10|100|1000 ] \
[ duplex half|full ] \
[ port tp|aui|bnc|mii|fibre ] \
...
...


FreeBSD case http://www.cyberciti.biz/faq/howto-configure-freebsd-full-half-duplex-speed/

Note About Duplex Settings

By default, Linux NICs negotiate their speed and duplex settings with the switch. This is done by exchanging electronic signals called Fast Link Pulses (FLP). When the speed and duplex are forced to a particular setting the FLPs are not sent. When a NIC is in auto-negotiation mode and detects a healthy, viable link but receives no FLPs, it errs on the side of caution and sets its duplex to half-duplex and sometimes it will also set its speed to the lowest configurable value. It is therefore possible to force a switch port to 100 Mbps full duplex, but have the auto-negotiating server NIC set itself to 100Mbps half-duplex which will result in errors. The same is true for the switch if the switch port is set to auto-negotiate and server NIC is set to 100 Mbps full duplex. It is best to either force both the switch port and server NIC to either auto-negotiate or the same forced speed and duplex values.
lik
Founder
Founder
 
Posts: 497
Joined: Wed Dec 15, 2010 3:21 am


Return to Linux specific

 


  • Related topics
    Replies
    Views
    Last post