Hi Gary,
I was responding to Rob's message, where he referred to the client 
system as the "from" system, and the server system as the "to" system.
I don't know what "host" and "target" mean in your message.  To me, they 
are even more ambiguous as "from" and "to".  All systems can be referred 
to as hosts.
This thread originated with a poster saying he wanted to secure his 
system by locking down the "SBMRMTCMD" command.  In other words, he's 
protecting the server system by restricting access to the client-side 
command.
The problem with that is that it only restricts one command on one 
computer.  Anyone can either write, buy, or download a free tool to 
submit the same command without using the IBM-supplied SBMRMTCMD. 
Likewise, they can submit the command from another computer. In all 
cases, bypassing the security placed on the SBMRMTCMD parameter.
I suppose in a very small, tightly controlled, environment, it might be 
possible to control every computer that can submit a command, and every 
program on that computer.  But in that case, there's a lot more going on 
than simple securing access to SBMRMTCMD, there's also the network 
security and physical security that prevents attackers from bringing in 
their own computer, or connecting from outside.  Even with that level of 
physical and network security, it's easy for someone down the road to 
forget to enforce some aspect of it, and bring in a smartphone or 
computer that hasn't been secured. Or allow someone to write/install a 
program that could run a remote command. It's not a good way to run 
security!!
The proper way to secure remote commands is on the server-side.  The 
TCP/IP service that receives the remote command could have exit programs 
to restrict which commands can be run.  Or a userid/password (or with 
newer tools, a digital key or certificate) can be used to enforce the 
operating system's object-level security.  Once this security is 
enforced on the server-side, it doesn't matter how the client-side 
program is secured, or which program they run, or which computer the 
request comes from, because the server will only run the commands that 
are allowed.
The fact that people on this list (besides Chuck and Rob) seem to be 
advocating securing the SBMRMTCMD or RUNRMTCMD command is just silly.
On 5/17/2012 2:12 PM, Monnier, Gary wrote:
Scott,
By "to" machine do you mean the target system or the host system?
Gary Monnier
As an Amazon Associate we earn from qualifying purchases.