We are using Mercator Open Edition+PeopleSoft on AIX and have been very happy with it so far. This is a migration effort from and older MVS TSI product to Mercator. Nonetheless, TSI (the vendor) has told us that the NT version runs just as fast or faster (than the UNIX version) and cost less. Another benefit is that Mercator is written first for NT and then for other platforms. That means that the NT version is slightly ahead. If you are going to look at Mercator, take a look at Mercator EC and/or Mercator Open Edition. The only thing I find a pain in Mercator is debugging using the trace file. Jonathan Showalter Omaha, NE |--------+-----------------------> | | Anthony | | | Beecher | | | <anthonyb@BUY| | | .COM> | | | | | | 01/10/2000 | | | 01:14 PM | | | Please | | | respond to | | | Anthony | | | Beecher | | | | |--------+-----------------------> >----------------------------------------------------------------------------| | | | To: [EMAIL PROTECTED] | | cc: (bcc: Jonathan Showalter/MutualOMA) | | Subject: Re: EDI & ANY to XML translator - opinions wanted | >----------------------------------------------------------------------------| I had this thought, too - since the Mercator pricing is per CPU, it seems you get more and more for you money as the chips get faster and faster. I believe you were addressing Gentran however. I can offer some tips for more speed: 1. I/O seems to be a limitation, use a RAID array for disk speed. 2. I think the database is a bottleneck. I put the database on its own spearate physical disk and saw a nice improvement- less contention with the Gentran application's file activity. 3. If you don't use a distributed system, you might change the paths from network paths to local drive paths (i.e. change '\\Gentran' to 'e:\') (I didn't benchmark this, but it seemed a little faster for not running everything through network layer code) Anthony -----Original Message----- From: Glenn Thompson [mailto:[EMAIL PROTECTED]] Sent: Monday, January 10, 2000 10:41 AM To: [EMAIL PROTECTED] Subject: Re: EDI & ANY to XML translator - opinions wanted Brian, You seem very knowledgable about this. What would moving up to one of the new 733-750 MHz machines do. Would that provide any relief or would it be negligible? I am assuming of course that such a machine is not being used now. Let's just assume a 400 Mhz CPU. brian lehrhoff wrote: > Mentor and Director are the same - small files flow, large files > choke. The only way to achieve throughput on the PC platform is to > feed a steady stream of small files. I can't tell you the optimum > size for Director, but Mentor seems to top out at 700-800K per flat > file. The Unix flavor of Mentor does not have this "problem". > > Mercator would be a good choice for your solution, but keep in mind > that Mercator will read and validate your entire input file before > writing. Make sure your production machine has plenty of memory > (fastest) or disk (slower) for the work files. Same caveat prevails - > smaller input files run faster. > > Outsourcing? Glad to talk to you about it. Also glad to give you a > straight answer to your question :) > > Anthony Beecher wrote: > -- Glenn Thompson Programmer/Analyst American Trouser, Inc. [EMAIL PROTECTED] ======================================================================= To signoff the EDI-L list, mailto:[EMAIL PROTECTED] To subscribe, mailto:[EMAIL PROTECTED] To contact the list owner: mailto:[EMAIL PROTECTED] Archives at http://www.mail-archive.com/edi-l%40listserv.ucop.edu/Title: RE: EDI & ANY to XML translator - opinions wanted
I had this thought, too - since the Mercator pricing is per CPU, it seems you get more and more for you money as the chips get faster and faster.
I believe you were addressing Gentran however. I can offer some tips for more speed:
1. I/O seems to be a limitation, use a RAID array for disk speed.
2. I think the database is a bottleneck. I put the database on its own spearate physical disk and saw a nice improvement- less contention with the Gentran application's file activity.
3. If you don't use a distributed system, you might change the paths from network paths to local drive paths (i.e. change '\\Gentran' to 'e:\') (I didn't benchmark this, but it seemed a little faster for not running everything through network layer code)
Anthony
-----Original Message-----
From: Glenn Thompson [mailto:[EMAIL PROTECTED]]
Sent: Monday, January 10, 2000 10:41 AM
To: [EMAIL PROTECTED]
Subject: Re: EDI & ANY to XML translator - opinions wanted
Brian,
You seem very knowledgable about this. What would moving up to one of
the new 733-750 MHz machines do. Would that provide any relief or would
it be negligible? I am assuming of course that such a machine is not
being used now. Let's just assume a 400 Mhz CPU.
brian lehrhoff wrote:
> Mentor and Director are the same - small files flow, large files
> choke. The only way to achieve throughput on the PC platform is to
> feed a steady stream of small files. I can't tell you the optimum
> size for Director, but Mentor seems to top out at 700-800K per flat
> file. The Unix flavor of Mentor does not have this "problem".
>
> Mercator would be a good choice for your solution, but keep in mind
> that Mercator will read and validate your entire input file before
> writing. Make sure your production machine has plenty of memory
> (fastest) or disk (slower) for the work files. Same caveat prevails -
> smaller input files run faster.
>
> Outsourcing? Glad to talk to you about it. Also glad to give you a
> straight answer to your question :)
>
> Anthony Beecher wrote:
>
--
Glenn Thompson
Programmer/Analyst
American Trouser, Inc.
[EMAIL PROTECTED]
=======================================================================
To signoff the EDI-L list, mailto:[EMAIL PROTECTED]
To subscribe, mailto:[EMAIL PROTECTED]
To contact the list owner: mailto:[EMAIL PROTECTED]
Archives at http://www.mail-archive.com/edi-l%40listserv.ucop.edu/
