I think the best way to handle this is with separate deployments (rather than separate containers): a primary deployment and N research deployments. All data collection, trading, and connection to IB would happen on the primary deployment. Research deployments wouldn't run launchpad
, ibg
, or the blotter
. Databases would be synced to research deployments via S3. Different research deployments could load code from the same Git repo or from separate repos (and similarly with pushing to Git). Separate deployments would have separate URLs, separate houston
containers, and therefore separate Basic Auth credentials. The sys admin controls access by virtue of storing the several Docker Compose files (with associated Houston credentials) on their own machine.
The current terms of service allow running two separate deployments under one license key (one "production" deployment and one "development" deployment, perhaps better labeled "primary" and "research"). So you could run a separate research deployment now; everything I described above is already possible.
To accommodate funds that may want multiple research deployments in addition to the primary deployment, I think it would make sense to modify the license terms such that an additional research deployment is included for each increment of the account limit on the Supersonic plan. In other words, the base Supersonic plan (like all plans) gives you 1 primary and 1 research deployment; an account limit of $2M USD equivalent would give you 1 primary and 2 research deployments, $3M would give you 1primary and 3 research deployments, and so on. That should be plenty of deployments for any given AUM.