Tab navigation
- Overview
- What's new
- Products
- Performance - selected tab,
Fibre Channel Protocol (FCP) supports access to Small Computer System Interface (SCSI) peripheral devices in the z/VM, z/VSE, and Linux on System z environments. This is in contrast to FICON, the protocol which supports access to Extended Count Key Data (ECKD) peripheral devices.
FCP utilizes the Queued Direct Input/Output (QDIO) architecture, in similar fashion to the Open Systems Adapter (OSA). Rather than using traditional channel programs, QDIO enables direct memory-to-memory transfers between the host and the channel.
FCP performance metrics
When a FICON channel is configured as CHPID type FCP, I/O information is made available relating to FCP performance (latencies) and FCP channel usage. Linux on System z can extract the hardware statistics (time spent in the fabric and time spent on the channel) to assist with the preparation of graphics and help analyze the performance and usage of FCP channels. FCP performance metrics applies to the FICON Express8, FICON Express4 and FICON Express2 features (CHPID type FCP) on System z10 and System z9 servers with Linux on System z.
FCP performance increase for small block sizes
With the introduction of System z10, the Fibre Channel Protocol (FCP) Licensed Internal Code was modified to help provide increased I/O operations per second for small block sizes.
Performance results
The results shown were achieved in a laboratory environment using one channel configured as CHPID type FCP with no other processing occurring and do not represent actual field measurements.
I/Os per second - Read/writes/mix
4k block size, channel 100% utilized

The table above shows FCP performance characteristics on System z start - I/Os. There was a 90% increase on System z10 compared to System z9 using FICON Express4 and a 40% increase comparing FICON Express8 (FEx8) to FICON Express4 (FEx4) on System z10.
System z10 megabytes per second (half-duplex)
Large sequential reads versus writes

The table above shows FCP megabytes per second throughput on System z10 with half-duplex data transfers. There was an 80 to 100% increase comparing FICON Express8 (FEx8) to FICON Express4 (FEx4).
System z10 megabytes per second (full-duplex)
Large sequential read/write mix

The table above shows FCP megabytes per second throughput on System z10 full-duplex data transfers. There was a 40% increase comparing FICON Express8 to FICON Express4.
