You need to use a character set that includes the Microsoft Windows "special" characters. Or you could encode them using XML equivalents, and process the data as if it were an XML document.
Here's a useful technique. In the Transformer stage, select all the Char columns that are being mapped to VarChar, right click and choose Derivation Substitution. In the formula field type Trim($1) - this will apply the Trim() function to all the selected derivations.
If it's truly binary then you could compare it against the value of Char(01).
Tip: evaluate the Char() function once in a stage variable and compare against that value. In ASCII, Char(01) is Ctrl-A. If this doesn't work, then it's not actually a binary value that you're looking at.
If the hostname of the services tier can be resolved, what about the hostname of the engine tier?
If these are the same, have you confirmed that all required services are running (server1 on the services tier, ASBNode and Logging agents and DataStage RPC daemon - at least - on the engine tier)?
I think the first thing you need to do is to find an LDAP administrator on site from whom you can obtain these details. A Java practitioner would be a good find also. Your lack of experience is telling; learn from those experts.
To put it another way, because the file is fixed-width, the spaces are part of the data. You must import them, and transform (trim) them subsequently. In DataStage parallel jobs the philosophy is "one task, one stage". The CFF stage reads the CFF, the Transformer (or Modify) stage modifies...
Can that user's PC resolve the host name to an IP address (try it with PING from a CMD shell)? If not, edit the hosts file on that PC so that it can resolve the host name.