Installing the panagenda MarvelClient databases allows for the following types of installation.
In this setup, both databases (Config and Analyze) are installed on a local IBM Notes client.
Whilst a local installation is well suited for a quick evaluation, it does not allow for testing with multiple users.
In this setup, both databases (Config and Analyze) are installed on one dedicated IBM Domino server.
Whilst we do not recommend to use a dedicated server for lack of potentially needed scalability, you can “hard-code” MarvelClient to use one particular server as opposed to each user’s mailserver.
If used for production purposes, all users and/or test servers must have a good connection to this server. The server also needs to be able to process the maximum number of concurrent sessions at peak times – this number depends on the number of users that potentially log on during peak working hours and the session duration on the Domino server.
Recommended for environments up to a few thousand users: Load-balanced/Production
In this setup, the Config and Analyze databases are replicated across all mail servers, where each Analyze database contains *all data from all users* across all mail servers.
The advantage of this "non-hardcoded" installation type is, that any change at the server level (such as moving a user from one Mailserver to another, or adding new servers) doesn’t affect MarvelClient: In notes.ini; the corresponding MC_DB= entry refers to "%notes_homeserver%!!YOURDIRECTORY\YOURCONFIGDATABASENAME.nsf" , which means that there is no hard-coded servername.
Note: %notes_homeserver% is dynamically resolved from a user's then current location document.
Also note: You must make sure that the Audit agent in the Analyze database only runs on ONE server, which then replicates audited documents back to all other servers (if each mail server were to run the audit agent, this would generate a large number of replication conflicts).
Recommended for all environments, especially for many (hundreds of) thousands of users: Hub/Spoke
Similar to the Load-balanced/Production type install, the Config and Analyze databases are replicated onto every mail server, too. However, in order to reduce the size of the Analyze database on each mail server to just the documents of each mailserver's respective user population, you choose for a "Hub and Spoke" setup: In this setup scenario, all Analyze databases are replicated together into one large Analyze database on a hub. In the Analyze database, the only hub server is granted the role “[Admin]”, whereas the mail servers are not. Thereby the Analyze databases on mail servers only store the analyze data of “their own users” and replicate it to the hub, where all data is gathered.
Note: As to avoid replication conflicts, you must only schedule the Audit All Agent in the Analyze database on EITHER the hub only OR all spokes excluding the hub.