Chapter 10. DB2 I/O Operations

The information shown here is extracted and modified from different sections of the DB2 UDB for OS/390 V6 Administration Guide, SC26-8957. This information is provided to give storage administrators an understanding of the I/O operations performed by DB2.

The two most important I/O operations performed by DB2 are the data read I/O and the log write I/O. The data read I/O has direct impact on the response time of any SQL query. The log write I/O has an important impact on online transaction response time. This chapter will describe the following I/O operations in detail:

•Data I/O: read and write accesses to DB2 table spaces and index spaces

•Log I/O: read and write accesses to the active log.

Other I/O operations are performed by DB2, like access to image copies, archive logs, and BSDSs. These I/O operations are not described in this chapter.

10.1 Avoiding I/O Operations

One of the basic principles of DB2 design is to avoid I/O operations if at all possible. As Roger Miller, DB2 Lead Architect, often says: "The best I/O is the

one which is avoided completely, the second best is the one that does not go to the disk - just to the cache."

DB2 tries to achieve this, using a hierarchy of buffer pools to keep data in memory. Modern disk devices complement this by having large caches which give an additional level of intermediate data storage. This storage hierarchy is illustrated in Figure 32 on page 104.

DB2 uses virtual buffer pools to store the data pages. The virtual buffer pools are optionally backed up by hiper pools. When data sharing is used, group buffer pools in the coupling facility store updated pages before these are cast out to disk. Pages in the group buffer pool can be accessed from any member of the data sharing group.

In addition to the caching done by DB2, the storage controller also uses a cache for data. The controller has algorithms to determine the convenience of pre-staging the data to the cache. For example, if several sequential reads are detected, the controller reads tracks ahead of the requests in order to to improve the cache hit ratio.

With DB2 V6, the hierarchy shown in Figure 32 is extended to include virtual buffer pools in data spaces. For more details, refer to DB2 UDB for OS/390 Version 6 Performance Topics, SG24-5351.

© Copyright IBM Corp. 1999

103

Page 124 Page 126
Page 125
Image 125
IBM 5695-DF1, 5655-DB2 manual DB2 I/O Operations, Avoiding I/O Operations, 103
Page 124 Page 126

5695-DF1, 5655-DB2 specifications

IBM 5655-DB2 and 5695-DF1 are significant components within the IBM software ecosystem, predominantly focusing on data management and integration solutions. These offerings cater primarily to enterprise environments that require robust database management systems and associated frameworks to maintain and manipulate data efficiently.

IBM 5655-DB2 is a well-known relational database management system (RDBMS) that excels in managing large volumes of structured data. Its architecture is designed to support high availability, scalability, and performance, crucial for businesses operating in today’s data-driven world. Some of its main features include advanced indexing capabilities, support for complex queries, and dynamic workload management. Additionally, it provides strong concurrency controls, which enable multiple users to access and manipulate data simultaneously without compromising data integrity.

One of the key characteristics of DB2 is its support for various data types, including JSON and XML, making it versatile for modern applications that generate data in diverse formats. It also features robust security mechanisms to protect sensitive data, aligning with compliance standards across industries. Integration with analytics tools further allows businesses to derive insights from their data, enhancing decision-making processes.

On the other hand, IBM 5695-DF1, also known as the InfoSphere DataStage, is a powerful data integration tool that facilitates the extraction, transformation, and loading (ETL) of data from various sources to target systems. It empowers organizations to streamline their data flows, ensuring that clean, consistent information is available for analysis and operational use. Key features of 5695-DF1 include a user-friendly graphical interface that enhances developer productivity and a rich set of connectors for numerous data sources, enabling seamless data integration.

DataStage also supports real-time data integration, allowing businesses to keep their data synchronized across multiple platforms. Its parallel processing capabilities dedicatedly optimize performance, enabling organizations to handle vast datasets efficiently. It incorporates data quality tools that help in validating and cleansing data before it is used for decision-making processes.

Both IBM 5655-DB2 and 5695-DF1 are part of a broader strategy to accommodate the evolving landscape of data management. Businesses leverage these technologies to enhance their data architectures, fostering agility and competitive advantage in their respective markets. Their integration capabilities, along with a focus on security and scalability, position them as vital assets in modern enterprise environments. Whether managing critical data within a database or ensuring seamless data flow across systems, these IBM offerings provide a comprehensive approach to handling complex data challenges.
Top Page Image Contents