Out of curiosity, does anyone know what protocol is used for file transfers by 
the ISPF Workstation Agent (WSA.EXE)?
 
Dave Salt

SimpList(tm) - try it; you'll get it! 

http://www.mackinney.com/products/program-development/simplist.html  




> Date: Wed, 19 Sep 2012 09:39:37 -0500
> From: [email protected]
> Subject: Re: Why File transfer through TSO IND$FILE is slower than TCP/IP FTP 
> ?
> To: [email protected]
> 
> On 09/19/2012 07:39 AM, McKown, John wrote:
> > ftp does not go through as many transitions. Basically, there is an IP 
> > connection to a server which write directly to a file. The only 
> > encode/decode is if you do an ASCII translation. With TSO IND$FILE, you 
> > must encode and decode all your data into a 3270 data stream. Which, due to 
> > TSO, is half duplex (like the old walkie talkies, only one side talks at a 
> > time). You then IP connect your TN3270 emulator to a TN3270 server which 
> > encodes the 3270 data stream. Also, then IND$FILE transfer is limited to 
> > the size of your TSO screen. I.e. send one screen of encoded data (at most 
> > 80x24, or 1920, bytes for a -2; 27*132==3564 for a -5); wait and decode the 
> > response (protocol overhead); wash, rinse, repeat until all data is 
> > transferred. The protocol and data translation (if any) overhead on ftp is 
> > much less.
> >
> > The above is just my SWAG based on some knowledge of how IND$FILE works 
> > internally.
> >
> TSO File Transfer was also slow when TN3270 and IP were out of the 
> picture and you just had a coax-attached PC 3270 emulator with only 
> VTAM, 3270 protocol, and 3274 controllers involved.  Considering the age 
> of IND$FILE on MVS, there is probably minimal buffering used for 
> reading/writing the file under TSO on MVS.  Depending on DASD block size 
> and latency times, Just failing to adequately buffer the DASD file I/O 
> could easily degrade the throughput by an order of magnitude.  That,  
> plus the inefficiencies of trying to kludge arbitrary data into 
> half-duplex 3270 data-stream protocol, which was designed and optimized 
> for sending formatted text fields, is probably sufficient to explain why 
> this has never been a good way to transfer data files unless they were 
> very small.
> 
> Given the typical product and product-version unique, tedious, and 
> therefore error-prone methods for specifying IND$FILE file attributes 
> and file sizes on various 3270 emulators, I also found that even for 
> transferring small files with non-standard attributes, success was 
> easier and sooner obtained with FTP and MVS-consistent "quote site" 
> attribute overrides than trying to re-discover how to specify the 
> correct attributes through the unique GUI interfaces of some emulator 
> product or to decipher/modify existing transfer profiles unique to a 
> specific PC instance of that product.  I ceased to use IND$FILE except 
> when no other option for file transfer was available and the experience 
> in those few cases was frequently unpleasant by comparison.
> 
> -- 
> Joel C. Ewing,    Bentonville, AR       [email protected]       
> 
> ----------------------------------------------------------------------
> For IBM-MAIN subscribe / signoff / archive access instructions,
> send email to [email protected] with the message: INFO IBM-MAIN
                                          
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to