Author |
Message
|
titus |
Posted: Mon Oct 21, 2002 6:14 am Post subject: Performance |
|
|
Novice
Joined: 14 Oct 2002 Posts: 23
|
I am having a fixed length message format.
From the following options which one is best for performance
1) Creating a MRM an load MRM parser for parsing the message
2) Treat the message as Blob . Inside compute do a subString to get this fields.
Experts opinions are accepted
Thanks in advance
titus:) |
|
Back to top |
|
 |
zpat |
Posted: Mon Oct 21, 2002 8:07 am Post subject: |
|
|
 Jedi Council
Joined: 19 May 2001 Posts: 5866 Location: UK
|
Why not try each method with a few thousand test messages and see which works fastest?
My money is on the BLOB. |
|
Back to top |
|
 |
Tibor |
Posted: Mon Oct 21, 2002 12:35 pm Post subject: Re: Performance |
|
|
 Grand Master
Joined: 20 May 2001 Posts: 1033 Location: Hungary
|
titus wrote: |
I am having a fixed length message format.
From the following options which one is best for performance
1) Creating a MRM an load MRM parser for parsing the message
2) Treat the message as Blob . Inside compute do a subString to get this fields. |
- BLOB to BLOB wo/conversion is the fastest
- BLOB to XML, MRM, etc is fine, but you are losing the flexibility
- BLOB to non-BLOB may be slower then an MRM to MRM, when you are working with a complex structure, especially on repeating compound elements.
Tibor |
|
Back to top |
|
 |
titus |
Posted: Mon Oct 21, 2002 1:44 pm Post subject: |
|
|
Novice
Joined: 14 Oct 2002 Posts: 23
|
Thanks for ur reply Tibor.
Could you please explain first scenario
"BLOB to BLOB wo/conversion is the fastest ".
This what i am doing .Input node i made it as a Blob message structure and in compute node i am getting same Blob message and cast into a String variable. Then i am using this string variable for substring operation. Is this is a right approach? |
|
Back to top |
|
 |
Tibor |
Posted: Tue Oct 22, 2002 6:46 am Post subject: |
|
|
 Grand Master
Joined: 20 May 2001 Posts: 1033 Location: Hungary
|
titus wrote: |
...Input node i made it as a Blob message structure and in compute node i am getting same Blob message and cast into a String variable. Then i am using this string variable for substring operation. Is this is a right approach? |
It depends. If your ESQL code is simple this may be fast. But I don't see the difference between the structure of input and output messages. When your task is only a cut-and-paste, you ever leave CASTing.
Tibor |
|
Back to top |
|
 |
|