HP StorageWorks Fabric OS 6.1.1 administrator guide (5697-0235, December 2009)

394 Configuring and monitoring FCIP extension services
fastwrite flows may be passed through the FCIP tunnel, but only if the FCIP fastwrite option is disabled on
the tunnel.
FCIP fastwrite and tape pipelining
When the FCIP link is the slowest part of the network, consider using FCIP fastwrite and tape pipelining.
Supported only in Fabric OS 5.2.x and later, FCIP fastwrite and tape pipelining are two features that
provide accelerated speeds for write I/O over FCIP tunnels in some configurations:
• FCIP fastwrite accelerates the SCSI write I/Os over FCIP.
• Tape pipelining accelerates SCSI write and read I/Os to sequential devices (such as tape drives) over
FCIP, which reduces the number of round-trip times needed to complete the I/O over the IP network and
speeds up the process. To use tape pipelining, you must also enable fastwrite.
• Both sides of an FCIP tunnel must have matching configurations for these features to work.
FCIP fastwrite and tape pipelining are enabled by turning them on during the tunnel configuration process.
They are enabled on a per-FCIP tunnel basis. See”Managing FCIP tunnels” on page 409 for details.
Constraints for FCIP fastwrite and tape pipelining
Consider the constraints described in Table 89 when configuring tunnels to use either of these features.
Table 89 Using FCIP fastwrite and tape pipelining
FCIP fastwrite Tape pipelining
Each GbE port supports up to 2048
simultaneous accelerated exchanges, which
means a total of 2048 simultaneous
exchanges combined for fastwrite and tape
pipelining.
Each GbE port supports up to 2048
simultaneous accelerated exchanges, which
means a total of 2048 simultaneous
exchanges combined for fastwrite and tape
pipelining.
Does not affect FICON traffic Does not affect FICON traffic
FCIP fastwrite and FC fastwrite are mutually
exclusive.
Tape pipelining uses FCIP fastwrite, not FC
fastwrite.
Does not support multiple equal-cost path
configurations (see ”FCIP fastwrite/tape
pipelining configurations” on page 395).
Does not support multiple equal-cost path
configurations or multiple non-equal-cost path
configurations (see ”FCIP fastwrite/tape
pipelining configurations” on page 395).
Class 3 traffic is accelerated with fastwrite. Class 3 traffic is accelerated between host
and sequential device.
With sequential devices (tape drives), there
are 1024 initiator-tape (IT) pairs per GbE
port, but 2048 initiator-tape-LUN (ITL) pairs
per GbE port. The ITL pairs are shared among
the IT pairs. For example:
Two ITL pairs for each IT pair as long as the
target has two LUNs.
If a target has 32 LUNs, 32 ITL pairs for IT
pairs. In this case, only 64 IT pairs are
associated with ITL pairs.
The rest of the IT pairs are not associated to
any ITL pairs, so no tape pipelining is
performed for those pairs. By default, only
fastwrite-based acceleration is performed on
the unassociated pairs.
Does not support multiple non-equal-cost path
between host and sequential device