Glossary

A metric which attempts to quantify the level or rate of turnover of the buffers in the buffer pool expressed as turns per hour. Values less than 10 tend to indicate a properly sized buffer pool while values over 10 increasingly indicate excessive buffer thrashing.

A metric which attempts to quantify the percentage of requests to write to a buffer which have to wait either for a locked buffer to be released so that it can be overwritten with a different data or index page or for an locked LRU queue to be released so that that queue's least recently used cache page can be overwritten. This is a measure of concurrency contention within the server.

An Informix term to describe disk files, partitions, or devices given to the IDS server to use for data storage.

An Informix term to describe logical groupings of chunks collected for use by a database or its tables.

Contiguous disk space reserved from a single chunk to be used to extend a partition.

This term is used ambiguously. Traditionally, in data processing terms, a fragment is a non-contiguous portion of disk which is part of a file. It is used in this way to describe Informix partitions with more than one extent.

It is also the official Informix term for the parts of a table or index which have been divided into separate partitions. A table with more than one partition is also called a fragmented table and each of its partitions is referred to as a fragment. It is also the keyword used to define each partition of a fragmented table or index.

The process of a partition becoming fragmented.

Also the process of defining or altering the intentional fragmentation scheme of a table or index.

A metric that attempts to quantify the percentage of queries that are required to wait for object locks to be released so that the query can itself acquire a lock on the object.

A metric that attempts to quantify the rate at which logical log buffers are filled. It is used to size the logical log buffers to balance minimizing performance impact and the risk of data loss from an unflushed buffer.

A metric that attempts to quantify the level of transaction data flowing through the IDS instance. It is primarily used to adjust the size and number of logical logs configured.

An Informix concept describing the ultimate unit of data storage. All Informix data is organized into pages. Prior to IDS 10.00 IDS page size was fixed at 2K on most platforms and 4K on Windows and AIX. Newer versions permit each dbspace, except the ROOTDB dbspace, to be created with a user specified page size in increments of 2K beginning with the platform's default page size.

The primary object unit in an Informix Dynamic Server database. Each nonfragmented table or index or each fragment of a fragmented table or index is a partition. Each partition is identified in the server by a partition header page which records the location and size of the extents which make up the partition's data and other descriptive details.

The name of a partition.

A metric that attempts to quantify the rate at which logical log buffers are filled. It is used to size the \physical log buffer to minimize the impact of physical log IO.

The percentage of page read requests that are satisfied by pages already existing in the buffer cache.

A metric which attempts to determine the percentage of pages read in anticipation of use which are actually used. Excessive read ahead causes buffer thrashing and reduces cache hit ratios.

The percentage of page write and rewrite requests that are satisfied by pages already existing in the buffer cache.