When you configure the DAH distribution modes, the main choice is whether to use mirror mode, or non-mirror mode. This choice has an impact on all the other choices that you make in the configuration.
You should use mirror mode only if all the child servers under the DAH return identical responses to identical actions. A common example of this is a set of Content servers that each have an index containing the whole data set.
In all other cases, you must use non-mirror mode. For example, you use this option for a set of Content servers that each have an index containing a part of the whole data set.
In mirror mode, you have multiple identical IDOL Server indexes. You can use the DIH to index all incoming documents into every child server. This method creates redundant copies of the index, which you can use to reduce the load on your child servers or to improve fault tolerance. See Distribution Methods.
In the general case, when all the child servers are the same, DAH needs only select one child server to send the action to. It then sends the server’s response back to the client, with little extra processing. You use the distribution methods to define how to select the child server to send the action to.
To use mirror mode, you set the MirrorMode configuration parameter to True. This option opens up a further configuration option, LoadBalanceMethod, but is generally less complex than non-mirror mode.
In non-mirror mode, you configure MirrorMode as False. Non-mirror mode is more intensive for the DAH. When DAH receives an action, it must:
Select a subset of the child servers to send the action to.
Modify the action as required.
Send the action to each of the child servers and receive the responses.
Merge the responses of the servers into its own response.
In non-mirror mode, you must also choose whether to use simple combinator mode, or virtual databases (VDBs):
In simple combinator mode, DAH sends actions to all available child servers, and passes through any database matching parameters without alteration.
Virtual databases are an advanced distribution mode, and require a large amount of configuration. You define a series of VDBs in the configuration file, and map them to databases in the child servers. See Virtual Databases.
Unless you need to define custom database mappings to child servers, or use advanced functionality such as multi-stage queries, Micro Focus strongly recommends using simple combinator mode. This option is the quickest and easiest to configure.
|
|