Coupling Facility Technology

Coupling Facility Technology
- best practices

Consider applying these key best practices in designing and building coupling facility technology for your parallel sysplex and data sharing IT environment:

  • Have at least two coupling facilities in order to provide high availability.
  • Use an external stand-alone CF for the Lock and SCA structures and an Internal CF(ICF) to provide protection against a double failure that results when a CF failure is accompanied by the failure of at least one z/OS image hosting a DB2 whose structures reside in the failed CF.
  • Duplex group buffer pool (GBP) structures to maximize availability with negligible duplexing performance overhead.
  • Place the lock and SCA structures in a coupling facility that does not contain group buffer pool structures.
  • Use dual IO channels between each CPC and a coupling facility to safeguard against channel failures.
  • Use a high performance external CF that matches the performance of the fastest processors in the sysplex and at worst not more than a one generation behind.
  • For a production sysplex employ 2 dedicated engines minimum per CF. A single-engine CF could be used in case of low CF request volume but add a second engine when the peak CF CPU utilization reaches 25%.
  • Size GBPs using CFSizer and carefully allocate the initial storage sizes for the lock and SCA structures. Be certain the allocated storage to one CF is large enough to support and accommodate the workloads in both CFs.
  • Set appropriate GBP thresholds.
  • Allow GBP structures to be altered automatically by invoking the Auto Alter function.
  • Be certain to allow the system to rebuild the DB2 structures into another CF during recovery from any failure.
for more info...

Planning for DB2® Data Sharing Deployment

Deploying DB2 data sharing is an extensive project that calls for the collaborative efforts and coordination of many IT groups including database administrators, systems programmers, system hardware, applications, and middleware and network staffs.

We, leveraging IBM resources, can work with and assist you in managing and staging the project and in achieving robust, optimal and cost effective data sharing implementation in a parallel sysplex environment.

Data sharing implemented in DB2 and Parallel Sysplex enables read and concurrent write access to a single DB2 shared data from different DB2 subsystems installed on multiple systems or CPCs (central processor complexes). Data sharing can be designed to build the foundation to achieve 24/7 application availability, superior scalability and DB2 capacity, excellent dynamic workload balancing and great configuration flexibility and systems management.

Key DB2 Data Sharing Deployment Plans:

Develop a roadmap and checklists that you need to execute and complete in order to be certain that all tasks required to implement data sharing in your production and test environments are included in your own project plans.

Capacity Planning

You need to make a careful analysis to determine the required capacity for the parallel sysplex that is running your DB2 data sharing workload. You should account for:
  • The variable DB2 data sharing CPU overhead
  • The fixed z/OS multi-system management overhead
  • The optimal number and capacity of the coupling facilities required to deliver your availability requirements and to accommodate your workload
  • The number and type of high speed IO channels to connect the coupling facilities to each processor in the sysplex
  • The memory of each coupling facility

Hardware Planning

In planning a Parallel Sysplex that supports DB2 data sharing, you need to provide for key hardware components:
  • Coupling facilities that provide the data sharing capabilities and structures for systems in the Sysplex
  • Sysplex Timers to provide time synchronization across all members of the sysplex
  • The processors that participate in data sharing should be capable of connecting to coupling facilities
  • High speed coupling facility IO channels to connect the processors to the coupling facilities and deliver the minimum response time

z/OS Software Planning

You are required to define policies and make z/OS changes to manage resources and workload to support D2 data sharing:
  • A CFRM (coupling facility resource management) policy
  • A SFM(system failure management) policy
  • Couple data sets to hold other policies including ARM, WLM, LOGR, connectivity, and availability policies and information about sysplex configuration and data sharing groups
  • Coupling facility IO channels definitions

DB2 Planning

Some key DB2 areas that should get a special attention and a high focus when planning your DB2 data sharing migration & implementation project are:
  • A naming convention for the data sharing group and its members
  • Developing implementation checklists based on best practice and customer experiences
  • Determining up front what are your business drivers for deploying data sharing and which workloads you need to run in the data sharing group will help you great deal to decide the best way to implement data sharing in your production environment. You may decide to enable data sharing for an existing DB2 subsystem as initiating member or build an entirely new data sharing group.
  • Controlling key performance factors in a data sharing environment including overhead, tuning GBPs & DB2 locking, monitoring tools, thread volumes and transaction run-time criteria. Understanding how locking and buffering are managed in a data sharing and how to minimize the associated overhead.
  • Modifying your existing DB2 recovery and operational procedures to incorporate the necessary changes introduced by data sharing. Set up the environment to handle component failure by properly defining the policies required to manage failure including the CFRM policy, SFM policy, and ARM policy.

 

Test Environment Planning

Key tasks in building your test data sharing group are:
  • Determining the configuration of your test parallel sysplex and data sharing group
  • Developing test strategy and building test cases
  • Generating the required test data
  • Deciding test’s success criteria

Disaster Recovery Planning

  • Update your current disaster recovery procedures to include tasks related to data sharing
  • Check that the hardware configuration at your disaster recovery site can support a Parallel Sysplex configuration
  • Schedule and execute a disaster recovery exercise on a test data sharing group

Network Planning

  • Assess DB2 data sharing impact on reducing the load on your network
  • Plan for connectivity of distributed applications to DB2 data sharing group (supported by DB2 Connect, JDBC and TCP/IP Sysplex Distributor, configured with Dynamic Virtual IP address (DVIPA))
  • Decide TCP/IP contact port and resync ports
  • If you need to implement VTAM generic resources, then get your network ready to use such resources

Project Management Planning

A project manager should be appointed to
  • provide an overall management
  • track task executions
  • coordinate the efforts of the many technical teams participating in the data sharing deployment program
  • ensure effective interactions among the teams involved including operations, DB2, z/OS, subsystems, networking and application development teams