CKD - Count Key Data

Enhanced Definition

Count Key Data (CKD) is a fundamental data recording format used by IBM mainframe direct access storage devices (DASD). It defines how data is physically organized on disk tracks, enabling efficient storage, retrieval, and management of records by the hardware.

Key Characteristics

- Track-Based Organization: Data is organized into records on a track, where each record consists of a Count area, an optional Key area, and a Data area.
- Count Area: Contains essential metadata for the record, including its physical address (cylinder, head, record number), record length, and flags. It's crucial for the hardware to locate and manage records.
- Key Area (Optional): If present, it stores a search key (e.g., a record identifier). The DASD hardware can perform direct searches based on this key, significantly speeding up access to specific records.
- Data Area: This section holds the actual user data for the record.
- Variable-Length Records: CKD supports variable-length records, allowing for efficient use of storage space by allocating only the necessary amount for each record.
- Hardware-Assisted I/O: The CKD format is tightly integrated with DASD hardware, enabling specialized channel commands (e.g., Search Key Equal, Read Data) to perform operations directly on the device, offloading CPU work.

Use Cases

- Storing Data Sets: The underlying format for virtually all z/OS data sets, including Physical Sequential (PS), Partitioned (PO/PDS/PDSE), and Direct Access (DA) data sets.
- System Volumes: Used for critical system components like the SYSRES volume, VTOC (Volume Table Of Contents), and system libraries.
- Database Storage: Provides the physical storage foundation for mainframe databases like DB2 for z/OS and IMS, which manage their own logical record structures on top of the CKD format.
- Indexed Sequential Access Method (ISAM): Historically, CKD's key area was integral to ISAM, allowing efficient access to records via an index. While ISAM is largely superseded by VSAM, the CKD structure remains.
- Temporary Storage: Used for temporary data sets (TEMP files) created during job execution for intermediate processing.

Related Concepts

CKD is intrinsically linked to DASD (Direct Access Storage Devices), serving as the logical organization format for the physical hardware. The VTOC (Volume Table Of Contents), which describes all data sets on a volume, is itself stored in CKD format, with each Data Set Control Block (DSCB) being a CKD record. Channel programs and CCWs (Channel Command Words) are used to issue I/O commands to the DASD controller to read, write, or search CKD records. Understanding CKD is fundamental to comprehending data set organization and how z/OS manages storage at a low level.

Best Practices:

Optimize Block Sizes: Choose appropriate block sizes for data sets to balance I/O efficiency and storage utilization. Larger blocks can reduce I/O operations but may increase memory requirements and internal fragmentation.
Efficient Key Design: When designing applications that utilize keyed access (e.g., for direct access files), ensure keys are unique, concise, and facilitate efficient hardware searches to minimize I/O and CPU overhead.
DASD Space Management: Regularly monitor and manage DASD space, understanding how CKD records are allocated and how fragmentation can impact performance. Utilize utilities for reorganization when necessary.
Performance Tuning: Analyze I/O statistics to identify bottlenecks related to CKD access. Consider data set placement on faster or less contended DASD volumes and optimize application access patterns.
Understand Physical Layout: For advanced performance tuning and problem diagnosis, have a basic understanding of how CKD maps to the physical cylinders and tracks of a DASD device.