[GitHub] rob05c commented on issue #2274: Profiles endpoint unordered

2018-05-18 Thread GitBox
rob05c commented on issue #2274: Profiles endpoint unordered
URL: 
https://github.com/apache/incubator-trafficcontrol/issues/2274#issuecomment-390222865
 
 
   I agree we should revisit `orderby` in API 2.0. However, currently, 
1.x/profiles doesn't support orderby, per 
https://traffic-control-cdn.readthedocs.io/en/latest/api/v12/profile.html
   
   I remain -1 on adding orderby as a new feature to endpoints that don't 
currently have it. Yes, our API is inconsistent, in a thousand ways. "Our API 
is inconsistent, therefore we should make it consistently bad" is not a good 
reason, either.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services


[GitHub] rob05c commented on issue #2274: Profiles endpoint unordered

2018-05-17 Thread GitBox
rob05c commented on issue #2274: Profiles endpoint unordered
URL: 
https://github.com/apache/incubator-trafficcontrol/issues/2274#issuecomment-390024452
 
 
   Just because something is common doesn't make it good. And yes, we have it 
on existing endpoints. But IMO we should be reducing unnecessary server load, 
not increasing it.
   
   In this particular case, Perl didn't "support" it, it was an incidental 
implementation. Right now, things like this aren't a big deal, because we have 
a very slow requests/second rate, due to changes being limited to operations. 
But as we move to Self Service, that's going to change, and load is going to 
start mattering more, as hundreds and even thousands of users are using Traffic 
Ops at once, and each user is changing configuration much more frequently.
   
   When that happens, there's a big difference in the TO server having to 
process 1000 concurrent requests and do the extra work to sort all of them at 
the same time, versus having all 1000 users sort their data client-side in the 
browser.
   
   Ordering one endpoint is small, but every decision we make like this affects 
the overall scalability of the application. We should make good decisions now, 
not putting unnecessary work on the server, so we don't have to do more work to 
refactor both the client and the server, when Self Service increases our load 
1000x.


This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


With regards,
Apache Git Services