Discussion (3)2
Log in or sign up to continue

This looks nice, and might get even more convenient if you'd add "shell" for interactive work and not only  API (you do have READ, WRITE, KILL entry points, but why not wrap them in the interactive shell?)

But here is the bigger concern - I've scanned thru your documentation, and haven't found any mention of security. You just open 5000,5001,5002 ports at each respective system, accept all incoming requests, not check any logins or passwords, or challenge phrases, or security tokens, and hope that there is no evil people in the world?

Eugene mentions OMI and ECP as traditional interconnection kinds of transport. None of them is quite secure, one may check docs for proof (as to ECP: http://docs.intersystems.com/latest/csp/docbook/DocBook.UI.Page.cls?KEY=... ).

ECP is defined as a basic service in contrast to resource based ones. Basic services can't efficiently use resource/roles/users security model by design. It's understood as their layer is usually too low to apply it.

Eugene's services can be dealt with the same precautions as other basic services, i.e. inside secured perimeter only. Port 500x can (and should) be closed on external firewall.