Re: http_post consumer API calls in a mass volume situation

We used to "load test" our network group's Cold Fusion servers with GETURI
on our IBM i back in the 90s. I wrote a program where you input a URL and
it would loop and submit 1000 or more jobs to QSYSNOMAX to make the
requests as close to simultaneous as possible,. We saved the start and end
times of each call and it just crippled it once it got to about 50
requests. They got mad but it also allowed the development of web apps to
be moved to my group on the IBM i (well, AS/400 back then).



On Thu, Mar 13, 2025 at 9:06 AM Charles Wilt <charles.wilt@xxxxxxxxx> wrote:

Given a "large volume", I'd probably stay away from http_post.

I've used Scott's HTTP API to send 1000's of requests per minute to an AWS
Lambda. Using a persistent connection (especially when dealing with
SSL/TLS) saves a considerable amount of time. In that case, the business
apps wrote to a queue and there was a set of 3 RPG jobs reading from the
queue and sending the data out.

I've also written a RPG handler program that used HTTP API to redirect RPG
writes to a web service...though that never ended up getting turned on in
production AFAIK.

Assuming your apps don't need to wait for a result, then the queue type
solution is likely your best bet.

You might also want to take a look at Apache Camel and/or Kafka.
Jesse Gorzinski has a number of presentations/articles about using them.

Good place to start
https://www.youtube.com/watch?v=8qVfw6KR2ww

HTH,
Charles



On Tue, Mar 11, 2025 at 11:06 AM Jay Vaughn <jeffersonvaughn@xxxxxxxxx>
wrote:

Let me start with just a pretty generic question and scenario.

If you had a need to send a large volume of http API consumer requests
from your native applications (and db2)... is it correct to think that
http_post may not exactly be the best choice as it is real time
synchronous...
If so, what are some other good options?

Or am I wrong? Could you expect http_post to hold up.

Pretty speculative scenario I realize - but imagine a level of volume

that

you may consider moderate to high even for your native applications to go
after db2 data locally - but instead, those "i/o"s are considering to be
replaced with http_post to another system.
As a change occurs in the application, that change is pushed over off
platform via http via the http_post...

can it work?
what are the considerations?
Also this would be with https (ssl authentication in play)

tia

jay
--
This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing

list

To post a message email: MIDRANGE-L@xxxxxxxxxxxxxxxxxx
To subscribe, unsubscribe, or change list options,
visit: https://lists.midrange.com/mailman/listinfo/midrange-l
or email: MIDRANGE-L-request@xxxxxxxxxxxxxxxxxx
Before posting, please take a moment to review the archives
at https://archive.midrange.com/midrange-l.

Please contact support@xxxxxxxxxxxxxxxxxxxx for any subscription related
questions.

--
This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing list
To post a message email: MIDRANGE-L@xxxxxxxxxxxxxxxxxx
To subscribe, unsubscribe, or change list options,
visit: https://lists.midrange.com/mailman/listinfo/midrange-l
or email: MIDRANGE-L-request@xxxxxxxxxxxxxxxxxx
Before posting, please take a moment to review the archives
at https://archive.midrange.com/midrange-l.

Please contact support@xxxxxxxxxxxxxxxxxxxx for any subscription related
questions.