The iSeries™ server supports Gigabit and 10 Gbps Ethernet, which requires the use of a 1 Gbps or 10 Gbps input/output adapter (IOA).
There are six IOAs that can be used to run Gigabit Ethernet, the 2760, 5701, and 5706 Unshielded Twisted Pair (UTP) adapters, which use copper wire cabling, and the 2743, 5700, and 5707 Gigabit Ethernet cards, which use fiber optics. The fiber optic 573A and 576A IOAs support 10 Gbps Ethernet. All of the adapter cards support Transmission Control Protocol/Internet Protocol (TCP/IP), 9000-byte jumbo frames, and the IEEE 802.3 standard.
i5/OS does not have integrated support for SNA over Gigabit or 10 Gbps Ethernet. Enterprise Extender or AnyNet is required to run SNA over Gigabit or 10 Gbps Ethernet.
The UTP cards have the capability of running 10BASE-T/100BASE-TX/1000BASE-T in full or half duplex mode. The recommended minimum cabling type for use between these cards and their link partner is Category 5e, which has copper twisted pair wiring. All four pairs are used when running Gigabit Ethernet. The cable and connectors must also be wired to TIA/EIA 568-B standard. Crossover cables are not supported.
The Gigabit fiber optic cards only support 1000BASE-SX in full duplex mode. The 10 Gigabit fiber optic cards only support 10Gbase-LR and 10Gbase-SR in full duplex mode. The 2743 and 576A use duplex SC connectors while the 5700, 5707, and 573A use duplex LC connectors. The recommended fiber to use with the 2743, 5700, 5707, and 573A IOAs is 50 or 62.5 micron MMF (multi mode) fiber. 9 micron SMF (single mode) fiber is recommended for the 576A.
IOA | Earliest release supported |
---|---|
2743 | V4R5 |
2760 | V5R1 |
5700 | V5R2 |
5701 | V5R2 |
5706 | V5R3 |
5707 | V5R3 |
573A | V5R3* |
576A | V5R3* |
Ethernet frames can contain anywhere from 64 to 9000 bytes of information. All of these cards support the largest, 9000-byte jumbo frame size. The amount of work for the adapter card to process each frame, regardless of size, is nearly the same. Therefore, you want to pack the most information possible into each frame. The final result will be to decrease processor utilization, leaving it more available for other application usage.