Extended Adaptive Cache Concepts

Explore Extended Adaptive Cache. Find information about planning, restrictions, and important considerations before you begin to use this tool.

Improve system performance with Extended Adaptive Cache, an advanced read cache technology that improves both the I/O subsystem and system response times by reducing the number of physical I/O requests that are read from disk. Extended Adaptive Cache not only improves the performance of database-read actions, but of all read actions. This includes read actions that are generated by other system components such as the Integrated xSeries® Server. It also works effectively in storage subsystems that have device parity protection or mirrored protection. Extended Adaptive Cache has proven to be highly effective on many types of workloads.

How the Extended Adaptive Cache works

Extended Adaptive Cache is integrated into the iSeries™ I/O subsystem. It operates at the disk subsystem controller level and does not affect the iSeries system processor. The storage I/O adapter manages the Extended Adaptive Cache by using a Read Cache Device (such as a solid state disk) to provide the cache memory.

Extended Adaptive Cache generates statistical information for the data, and then uses a mix of management strategies to determine which data to cache. The management of the cache is performed automatically within the I/O adapter and is designed to cache data by using a predictive algorithm. The algorithm considers how recently and how frequently the host has accessed a predetermined range of data.

The design of Extended Adaptive Cache was based on specific data management strategies of the iSeries server. Whether the disks are device parity protected, mirrored, or unprotected, the data stored on the disks has a tendency to occur in bands. This means that there are physically contiguous areas of disk storage where data is actively read, physically contiguous areas that are frequently written to, physically contiguous areas that are both actively read and written to, or physically contiguous areas of storage that are not frequently accessed.

This "banding" of data is accounted for in the Extended Adaptive Cache design. The goal is to cache bands characterized as read/write and read-only. A band that is characterized as write-only, while cached in the storage subsystem write cache, remains largely unaffected by Extended Adaptive Cache. Extended Adaptive Cache is also designed to not harm the performance of large blocks of data that are either sequentially written or sequentially read. In this instance, the pre-fetch capability of the disks, as well as other caches in the system, ensures a quick response time.

Related concepts
Device parity protection
Mirrored protection