Networking Enhancements

The following networking features and enhancements have been added to the Solaris 10 6/06 release.

IPv6 for IPFilter

This networking feature is new in the Solaris 10 6/06 release.

IPFilter for the Solaris OS has been updated to include IPv6 support. Packet filtering rules which include IPv6 addresses can be applied by using the ipf commands. IPv6 extension headers can be used to enable filtering. The IPv6 option has also been added to the ipfstat IPFilter statistics.

IPFilter can now be deployed in IPv6 networks to enhance security.

For further information see the ipf(1M) and ipfstat(1M) man pages. See also the System Administration Guide: IP Services.

UDP and TCP Performance Enhancement

This networking enhancement is new in the Solaris 10 6/06 release.

Performance of both the Transmission Control Protocol (TCP) and the User Datagram Protocol (UDP) has been enhanced in this release. The enhancements result in lower latency and higher throughput for both transmit performance and receive performance. Network applications yield better performance due to system performance improvements. This feature particularly benefits applications that heavily transmit and receive UDP packets or use TCP loopback connections.

For more information, see the ip(7P), tcp(7P), and udp(7P) man pages. See also the Solaris Tunable Parameters Reference Manual.

IP_NEXTHOP Socket Option

This networking enhancement is new in the Solaris 10 6/06 release.

The IP_NEXTHOP is a new IP-level socket option that specifies the address of the next hop for the traffic that originates from the socket. An application that has the IP_NEXTHOP option set bypasses the routing table lookups on the destination and sends packets directly to the specified onlink nexthop.

The thread that sets the IP_NEXTHOP option must have the PRIV_SYS_NET_CONFIG privilege.

TCP_INIT_CWND Socket Option

This networking enhancement is new in the Solaris 10 6/06 release.

The new TCP socket option, TCP_INIT_CWND, enables an application to override the settings in the initial TCP congestion window, as described in RFC 3390, “Increasing TCP's Initial Window.” By default, TCP sets the initial congestion window at connection set up time and after an idle period. (An idle period is when no traffic occurs between the two ends of the TCP connection. An application can use the TCP_INIT_CWND socket option to set the initial congestion window to a specified number of TCP segments. So the value of this new socket option is used both at the connection start time and after an idle period to set the initial congestion window. The process must have the PRIV_SYS_NET_CONFIG privilege if a number greater than that calculated by RFC 3390 must be specified.

For further information, see the tcp(7P) man page.