Re: Questions on "Calling PC Commands from RPG" article -- MIDRANGE-L

Obviously good points. I guess I see the beauty of the security in the Java
solution. It takes commands only from an iSeries data queue which, I
believe, can be locked down fairly securely. Locked down as in it'spossible to restrict who can send a DQE and perhaps even restrict the
application(s) that can do it.

I don't understand what you're saying here. You need a userid/passwordin order to even begin to know WHO you're locking it down to. You can'tlock something down to a user unless you've established that a user reallyis that user.

And that's what passwords do. (Granted, there are other means of doingso, Kerberos tickets, SSL Certificates, etc.)

And that's why a userid/password must be sent for it to be secure. Thefact that one solution uses a data queue really has no impact on thesecurity question at all.

In fact, if you're using a TCP/IP network, then there are IP datagramsthat the data is carried in. It doesn't matter if you use a data queue. (Itdoesn't matter if you use a green widget with olive oil dressing, for thatmatter) If it's travelling over TCP/IP, then someone else can create andsend the same TCP/IP packets and gain access to the command server.

Please someone tell me if I'm wrong on that.


You're wrong if you think that data queues make it more secure.

The possibility exists that the Java one that you've fouind is more securefor some other reason, though. (I wouldn't know.)

On the rexec side of the security equation, it seems less clear to me. I
know I've seen warnings about this in the past, but I've never pursued it
much prior to yesterday.

STRPCCMD solves the userid/password problem because it's part of anexisting 5250 connection that the user has already authenticated by (a)Having access to start a program on the Windows box in the first place,and (b) having successfully logged in to the iSeries.


So, in a way, it greatly simplifies the security situation.

On the other hand, the user has to trust the iSeries that he's connectedto. You'd certainly never want to connect to a web page and allow thatweb server to run arbitrary commands on your PC, why is it different with5250 and STRPCCMD? And the main reason that it's different is that userstypically know and trust the people who run their iSeries, but don't knowand trust the people who run a web server.

RUNRMTCMD/rexec does not make that same assumption that you trust theserver you're connected to. Instead, it asks for the userid and password.So -- you could look at it both ways. In one way, STRPCCMD is moresecure, in another, RUNRMTCMD is more secure.