With Extended Parallel Server, many coservers function together to form a single database server. These coservers usually have the same number of CPUs and equal amounts of memory and disk storage space. This uniformity makes it easier to administer all coservers from a single point.
The following features help you control and manage multiple coservers:
The database server has one centralized configuration file that applies to all coservers. To simplify the task of configuring multiple database server instances, Extended Parallel Server uses global configuration parameters for information that applies to all coservers. You specify these global configuration parameters once in the global section of the configuration file. The configuration file also contains a section for information that applies to individual coservers. For more information about global and coserver-specific configuration parameters, see Configuring Multiple Coservers.
A coserver group (referred to as a cogroup for convenience) is a subset of the coservers within the database server. You can create cogroups to simplify system and database administration. A coserver can be in more than one cogroup. For example, you might create the following cogroups.
Extended Parallel Server provides a system-defined cogroup called cogroup_all. The cogroup_all cogroup includes all of the coservers that your ONCONFIG configuration file defines.
Cogroups help you administer a large number of coservers. For more details see Initializing the Database Server.
A dbslice is a named set of dbspaces that usually spans multiple coservers. A dbslice is managed as a single storage object. For more information about dbslices, refer to Data Storage.
A logslice is a set of logical-log files that occupy a dbslice and are owned by multiple coservers, one logical-log file per dbspace. Logslices simplify the process of adding and deleting logical-log files by treating sets of them as single entities. For more information about logslices, see Logslices.
To facilitate administration, it is recommended that you use a centralized message-log file. To do so, place the message-log file in a file system that is accessible by all coservers. All coservers can then write their messages to this file. To identify the coserver that generated a given message, check the coserver number that begins each message.
Avoid placing message-log files in local file systems. That arrangement complicates administration of the database server by scattering log messages across multiple files. This arrangement also prevents you from viewing messages in sequence.
Extended Parallel Server provides centralized command-line and graphical monitoring utilities.
The onutil utility allows you to monitor and manage tables and other database objects across multiple coservers. For more information about onutil, refer to the utilities chapter in the IBM Informix: Administrator's Reference.
The xctl utility allows you to run coserver-specific utilities on multiple coservers and execute operating-system commands on multiple nodes. You can include a coserver-specific utility such as oninit, onlog, or onstat, or an operating-system command in an xctl command-line, along with syntax to designate the nodes or coservers on which the command is to run. For more information on xctl, refer to the utilities chapter in the IBM Informix: Administrator's Reference.
When a table is fragmented across coservers, each table fragment is owned by the coserver on which it resides. Only the owning coserver can update the system catalog tables that define the table fragment.
When a session executes an SQL statement that accesses data on a table fragment that is located on another coserver, the database server reads the system catalog tables from a participating coserver and stores them in structures that it can access more efficiently. These structures are created in the virtual portion of shared memory for use by all sessions on that coserver. These structures constitute the data-dictionary cache for a coserver.
Each coserver maintains its own data-dictionary cache. Each coserver can cache the data-dictionary information retrieved from other coservers that own a particular database definition. The database server ensures that each coserver accesses the most current system tables in its data-dictionary cache.
Only one INFORMIXDIR directory and one logical sqlhosts file exist across all coservers.
Home | [ Top of Page | Previous Page | Next Page | Contents | Index ]