Storage

z/OS Global Mirror Emulation vs. Persistent IU Pacing

3 Pages

The August/September 2011 z/Journal article, “Understanding Persistent IU Pacing (aka Extended Distance FICON)” (available at www.mainframezone.com/operating-systems/understanding-persistent-iu-pacing-aka-extended-distance-ficon) explained the basics of Information Unit (IU) pacing and the newer persistent IU pacing mechanism. The American National Standards Institute (ANSI) T11 FC-SB3 standard was amended (FC-SB3/AM1) in January 2007 to incorporate changes made with persistent IU pacing.

Persistent IU pacing is a set of enhancements designed to accommodate the improved performance of long-distance z/OS Global Mirror (zGM) implementations. The more commonly known term for these enhancements, which IBM announced in February 2008, is Extended Distance FICON. The previous article also explained how IU/persistent IU pacing functions work and touched briefly on their performance, noting that use of persistent IU pacing:

  • Doesn’t increase the distance support for Fibre Connection (FICON), but enhances the performance for a given distance when compared to the base IU pacing mechanism. (It does this by increasing the number of IUs that can be in flight between a channel and a control unit and allowing that increased number of IUs to persist for longer than one exchange.)
  • Doesn’t change the need for FICON directors or channel extension hardware used for multi-site Disaster Recovery/Business Continuity (DR/BC) and IT resiliency architectures
  • Could, in some cases, eliminate the need for zGM emulation technology (typically a feature license) on the channel extension hardware. This could reduce the overall cost of these architectures.
     

Channel Distance Extension Background

When introduced in 1990, Enterprise Systems Connection (ESCON) supported a maximum distance of 20 km. While it exploited the relatively new Fibre Channel technology, it employed circuit switching rather than the packet switching typical of today’s Fibre Channel implementations. This provided significantly improved physical connectivity, but retained the same one-at-a-time Channel Command Word (CCW) challenge-response/permission seeking logical connectivity that had been employed by parallel channels. ESCON channels only transmit a single CCW at a time to a storage subsystem. Once the subsystem (such as a DASD array) has completed its work with the CCW, it notifies the channel of this with a Channel-End/Device-End (CE/DE). The CE/DE presents the status of the CCW. If the status of the CCW was normal, the channel then transmits the next CCW of the channel program. As a result, ESCON performance is significantly reduced at extended distances by the multiple round-trip delays required to fetch the channel programs, a performance deficiency commonly referred to as ESCON droop. Figure 1 illustrates the ESCON CCW pattern.

By the late ’90s, the shortcomings of ESCON were driving a new technical solution. FICON evolved to address the technical limitations of ESCON in bandwidth, channel/device addressing, and distance. FICON channels rely on logical connectivity based on the notion of assumed completion rather than ESCON’s permission-seeking schema. The entire FICON channel program is transmitted to the storage subsystem at the start of an I/O operation. If there are more than 16 CCWs in the channel program, the remainder is transmitted in groups of eight until the channel program is completed. (The majority of channel programs have 16 or fewer CCWs.)

After the storage subsystem completes the entire channel program, it notifies the channel with a CE/DE. The numerous turnarounds present with the ESCON protocol are eliminated. These two changes allow a much higher percentage of the available bandwidth to be employed for data transmission and for significantly better performance over extended distances. Figure 2 illustrates the FICON CCW pattern.

 

IU/Persistent IU Pacing Review

IU pacing provides a load-sharing or fair-access mechanism for multiple, competing channel programs. While this facility yields desirable results, ensuring more predicable I/O response times on heavily loaded channels, it produces fewer optimal results for long-distance deployments. In these cases, increased link latencies can introduce dormant periods on the channel and its Wide Area Network (WAN) link. Dormant periods occur when delays waiting for anticipated command responses increase to the point where the pacing window prohibits the timely execution of CCWs that might otherwise be executed to ensure optimal performance.

3 Pages