High availability cluster

In a cluster, you gain scale-out, high availability, disaster recovery, and load-balancing capabilities. This technology is built right into the database and requires no additional hardware, software, or network infrastructure. In contrast to ER, in a cluster, you cannot control the granularity of data replication; the entire server's data is always replicated. Cluster and ER technologies, as this article shows later, can be used side by side.

A cluster consists of one primary server and a mix of three different types of secondary servers:
  • Shared Disk (SD) secondary server
  • High-Availability Data Replication (HDR) server
  • Remote Standalone (RS) secondary server

Servers must be the same HCL OneDB™ version and run on the same operating system, hardware, and disk architecture. Applicants can also insert, update, and delete data on secondary servers. Thus, all three secondary server types can be used to increase the capacity of your primary as well as balance the load.

Table 1, below, compares the server types. Table 2 lists some common ONCONFIG settings.

Table 1. Overview of the characteristics of cluster server types
Primary SD secondary HDR RS secondary
Primary purpose Increased capacity/scaleout High availability/hot standby Disaster recovery/remote backup
Maximum number of nodes per cluster One No limit One No limit
Possible type changes in one step SD secondary, HDR, standard Primary Primary, RS secondary, standard HDR, standard
Supported isolation levels All

Dirty read, Committed Read, Committed Read, Last Committed

Dirty read Dirty read
Table 2. ONCONFIG parameters common to cluster server types
Configuration parameter Server type Supported values Description
HA_ALIAS SD and RS secondary servers DBSERVERNAME or one of the DBSERVERALIAS name. Must be a network protocol.

Used to define the name by which a secondary server is known within a cluster. Gives you flexibility to use something other than DBSERVERNAME.

LOG_INDEX_BUILDS Primary, HDR, and RS secondary
  • 0- Disable index page logging
  • 1 - Enable index page logging
While creating a new index at primary server, Index Page Logging (IPL) writes index pages to logical log file. Enabling index page logging is required for RSS nodes and optional for HDR secondary server.
UPDATABLE_SECONDARY HDR, SD and RS secondary servers Primary Primary, RS secondary, standard
Supported isolation levels All
  • 0- Disable write functionality at secondary server
  • >=1 - Enable write functionality at secondary server
Determines the number of network connections used between the primary and secondary server to support data updates.
TEMPTAB_NOLOG Primary, HDR, SD, and RS secondary servers
  • 0- Create temporary tables with logging enabled by default
  • 1 - Create temporary tables without logging.

Required to be set to 1 at HDR, RSS, and SD secondary servers. Controls the default logging mode

for explicit temporary tables created with CREATE TEMP TABLE or SELECT INTO TEMP statements. Secondary servers must not have logged temporary tables.