Hi, there are many variables involved. 1) For example based on you expected traffic amount what is the expected time distribution of the service? Some approx. avg. calculations (which may differ from live traffic):
TOT_SMS | TOT_HOURS | SMSxHOUR | SMSxMIN | SMSxSEC 1000000 | 24 | 41666 | 694.4 | 11.6 1000000 | 20 | 50000 | 833.3 | 13.9 1000000 | 16 | 62500 | 1041.7 | 17.4 1000000 | 12 | 83333 | 1388.9 | 23.1 1000000 | 08 | 125000 | 2083.3 | 34.7 1000000 | 04 | 250000 | 4166.7 | 69.4 However those figures depend on your smpp connection(s) throughput. As for kannel itself there is no problem to process (send/receive) at those speeds. 2) Depending on the type of services, their response time, their distribution among shortcodes (if you have more than one shortcode and/or more than one operator) multiple smsboxes/sqlboxes may help. 3) To be avoided are large queues on bearerbox (when possible) If you care/can/want to share more specific details, we may ( or may not :) ) give different other advices. However the best way to test your system is to perform some benchmarks with smpp simulators (using the real actual sms-services) and check how your system behave with different configuration types. Most of the times kannel is NOT the bottleneck. php/perl/shell... services, sql queries, etc.. ,if not properly implemented, may "kill" your system performance. Br, Rinor On 06/15/2013 08:36 AM, Eric Beda wrote: > Hi guys, > > I have a setup where by I am connected via smpp to mobile operator, > receive messages via sms-service get-url, process it with php and send > replies via sqlbox (inserting into send_sms) > > This setup goes all fine and well during testing ( low traffic ) phase > except for a few niggles (not the point of this email). But I expect > to get huge amounts of traffic approx 1 million sms's a day, so my > question is what is the most efficient/best practice to setup kannel ?? > > Cheers
