This page is intended for discussion of the Trackwork Server with a focus on issues and concerns of scalability, performance, maintenance and day to day operations.

After a few weeks of testing with the Vasco Road/Teichert job we are closing in on 3 million GPS data points all in one database table. It makes me wonder how the product will scale if we have several of these jobs running in parallel on the system. Since all the GPS data lives in the database it would be hard if not impossible to rebuild the DB if it were to become corrupt. We could try and fallback to a backup (are we running backups?).

If we re-implement the server I would think moving to a system where the GPS data was stored external to the database might not be a bad idea. In fact, having all the data written to the filesystem first as primary storage then inserted into the database for application access might be a better solution. This would allow us to re-build the database from raw data at any time.

(this may not be the correct spot for this)

1. 64 bit compatibility - Initial testing shows TMM lib doesn't work on 64 bit Vista (UNCONFIRMED, may be operator error)
   1. Verify latest version
   2. Check to see if Vista has a 32 bit compatibility flag
   3. If confirmed, evaluate fix option
      1. Trackwork server (tough)
      2. Client library settings (32 bit compatibility mode?)
      3. Installer options? (if Vista has compatibility mode)

1. Unit Tests
   1. Full suite of unit tests to support regression testing
   2. App to simulate a TMM for unit testing, error injection and load testing.

Basic administration tasks are not addressed by the C implementation. A Java administration tool has been written to address a few of the basic tasks such as customer creation, removal and module configuration and assignment. Following is a list of items that are not covered or require manual configuration:

1. Project archiving (Backup)
2. Server version upgrade per customer
3. Load balancing/moving customer data to new servers
4. Firewall port allocation and configuration
5. More robust logging
   1. Log for each TMM per customer
   2. Log file rolling (log files are currently unbounded)
   3. Improved log messages