The HP NC360T is a dual-port gigabit NIC manufactured by Hewlett Packard. It runs the Intel 82571EB chipset and connects using PCI Express x4.
- Link Types: 10/100/1000 base-T, both full and half duplex
- Connection: PCIe x4*
- Power Consumption: 4.25W
- RX Buffer: 4096kB
- TX Buffer: 4096kB
- IEEE Standards Support: 802.3, 802.3u, 802.3x, 802.3ab, 802.3ad, 802.1p, 802.1Q
- Supports TCP/IP Checksum and Segmentation Offloading
- OS Support: Linux, Windows, BSD
*The card is wired for PCIe x4, however the Intel 82571EB chipset supports PCIe x1 as well. It is possible to cut off the excess pins on the PCIe connector to make the card fit in an x1 slot.
The NC360T supports all types of link bonding using the ifenslave package under Linux.
HP does not offer support for the NC360T for systems running OSes other than Windows Server. However, despite being an HP product, the NC360T can use the Intel PRO/1000 PT adapter driver. This driver is continually updated by Intel and supports up to and including Windows 8.1. This driver contains all the necessary functionality to bond the NIC into 1 of 5 different configurations.
The NC360T runs 2 BIOSes, capable of running independently, which can be entered using Ctrl-S on POST. If the Windows driver is installed, the POST timeout can be changed from its default of 2 seconds to 0, 2, 3, or 5 seconds. The card contains its own PXE stack and each port can boot independently.
The NC360T can be bonded like any other card with ifenslave. It does not use the maximum RX and TX buffers by default. To enable them to help avoid overruns, add the following 2 lines to
/etc/network/interfaces for each port of the NIC:
auto eth0 iface eth0 inet manual bond-master bond0 bond-primary eth0 eth1 pre-up /sbin/ifconfig $IFACE mtu 9000 post-up /sbin/ethtool -G $IFACE rx 4096 post-up /sbin/ethtool -G $IFACE tx 4096
After installing the driver, the card can be managed by entering the device properties of either port. This can be done using the Network and Sharing Centre, or the Device Manager. Begin by creating a team with both cards as members. The type of the team is determined by the use case, but often "ALB load balancing" is preferred as it requires no switch support. Once the team is created, a new virtual NIC is created and used by the OS. The TX and RX buffer size can be forced to 4096kB for each port from the same settings panel.