TM1 Server overview

The IBM® TM1® Server manages access to the TM1 data directory for TM1 clients.

The following figure illustrates the operations of a remote TM1 Server server. These operations are explained in the text that follows.

Arrows show flow of information when the server is started, running, retrieving cubes, or shutting down.
  1. When the TM1 server is started, all TM1 data is loaded from the TM1 data directory into RAM on the server machine. At the same time, the server opens a new transactional log file called tm1s.log in the data directory. After the cubes are loaded, the remote server is available.

    The remote TM1 Server registers itself with one or more Admin Servers so that clients can connect to the remote TM1 Server. Client applications contact Admin Servers to locate available TM1 Servers. The clients log into the TM1 Servers whose data they want to access.

  2. While the TM1 server is running, all cube data resides in RAM. All edits received from TM1 clients are stored in a transaction log file named tm1s.log.

    As new values are received from clients, the TM1 Serverwrites the records to the tm1s.log file, keeping track of every data change, including the date and time the edit occurred, and the ID of the client who made the edit.

  3. TM1 clients retrieve cube values from the server. Clients also send edits to cube values to the TM1 server. As the server calculates new values in response to client requests, the server stores them in memory, increasing the amount of memory used by the server.
  4. When the TM1 server is shut down, or when an explicit Save Data command is issued, any changes to cube values are written from the transactional log file to the data directory.

    All records in the tm1s.log file are saved to disk, and the transaction log file is renamed by appending a date/time stamp to it. The tm1s.log file is saved in the server's data directory to back out data transactions. For details, see the topic Backing out transactions using the transaction log in the TM1 Operations documentation.

    If the server is intentionally shut down without saving the changes, the log file is saved with a time/date stamp and the extension is changed to .rej. You can process the tm1syyyymmddhhmmss.rej file through TurboIntegrator to recover the transactions.

    To save all changes to the data on a TM1 Server at any time without shutting down the server, right-click a server in Server Explorer and click Save Data. All records in the tm1s.log file are immediately written to disk, the transaction log file is renamed by appending a date/time stamp to it, and a new tm1s.log file is created to accept any subsequent edits to cube values.

    Any changes to the metadata, such as dimension definitions and cube definitions, are immediately saved to disk. The changes to the metadata are not written to the transaction log file.