Data storage - Fluid Topics - 3.10

Install Fluid Topics on Multiple Servers

Operating system
Debian
Category
Technical Notes
Audience
public
Version
3.10

Fluid Topics data are stored in different locations and databases, depending on their types.

The following diagram shows an architecture with the Indexing environment and Back Office located on two different servers, coupled with one front, and their corresponding database locations:

32491

The following diagram shows an architecture with the Indexing environment and Back Office located on two different servers, coupled with two fronts, and their corresponding database locations:

32490

On the Front Environment (Web server + Reply environment)

  • One MongoDB database for the documentary content (books, unstructured documents, and so on). These data are used for content access.
  • One proprietary database for the search indexes. These data provide the search results.
  • At least one MongoDB database named fluid-topics-$SERVICE_ID-$SERVICE_STATUS for tenant-related data:
    • Bookmarks
    • Indexing reports
    • Enrich configuration
    • Features toggling configuration
    • Homepage configuration
    • Message configuration (labels)
    • Portal theme configuration
    • Personal Books (shared books, personal topics, and so on)
    • KHUB publishing report
    • Saved searches
    • Fluid Topics user sessions
    • KHUB sources
    • User information (credentials of internal users, their groups, and so on)

As tenant-related content is stored in the MongoDB databases, it is recommended to have several MongoDB instances in case one would crash. It will prevent from any data loss.

The tenant-related database can be separated from the Front environment. If the tenant-related database is not located on the Front environment but on a separate server, then it is necessary to indicate the database location in the mongo.conf.json file on the Front server.

The following diagram shows another possible architecture where the tenant-related database is separated from the Front environment and located in a dedicated server:

32491324913249132491

On the Indexing environment

  • One MongoDB database for documentary content (books, unstructured documents, and so on). These Data are used for the actual content streamed when accessing the Reader page.
  • One proprietary database for the search indexes. These data provide the search results.
  • One MongoDB database for Processing Pipelines (PaF) reports.

In the same way as for tenant-related database, the documentary and PaF databases of the Indexing environment can be separated.

On the Back Office Environment

  • One MongoDB database for the Administration Center
  • One H2 database for the Back Office users.
  • One Cassandra database for the Analytics.