Currently a dedicated front end server is being used for each Galaxy backend. This is unnecessary and requires that users go to a different url in the event of an outage. The front-end should scan through a list of available backends until it finds one that responds. If the primary backend is down it can notify the user that their data may be unavailable due to the outage. It may also be beneficial to allow users to select a backend in the event that they want to locate their data.
The difficulty with this is that the users API key needs to be replicated across all backends, or at least a mapping needs to be stored elsewhere.
Currently a dedicated front end server is being used for each Galaxy backend. This is unnecessary and requires that users go to a different url in the event of an outage. The front-end should scan through a list of available backends until it finds one that responds. If the primary backend is down it can notify the user that their data may be unavailable due to the outage. It may also be beneficial to allow users to select a backend in the event that they want to locate their data.
The difficulty with this is that the users API key needs to be replicated across all backends, or at least a mapping needs to be stored elsewhere.