As the load increases because of more users or traffic you should optimize
the way the database is used. Unfortunately because of different patterns of
usage between installations is not possible to apply one optimization with
the same cost.

The 'normal' platform based around a single SIP Proxy machine has been designed for handling traffic from about 30,000 subscribers.

Some calls may remain in the radius database in an un-normalized state. They may show in CDRtool as "in progress". 

 

These calls could not be normalized automatically by the system because there was no indication when the call has stopped as no BYE has been received and no media has been exchanged after the initial INVITE.

 

You may chose to delete such calls based on your own policy. 

 

Voicemail boxes are stored on the asterisk machine as directories. Extended 3 Linux file system has a limitation of how many directories can be created on the same level. To be able to create more than 30K directories you must setup a separate file-system using ReiserFS instead of ext3 for the storage of voicemail boxes and mount it to /var/spool/asterisk/voicemail

Try to reproduce the problem. Make a test call. Find the the call in CDRTool and /var/log/syslog. In CDRTool save the query. Provide the name of the saved query to AG Projects support. Even better, narrow the search to find a single call (use the SIP Session ID in the search screen), this helps pinpoint the problem and saves time when requesting support to AG Projects.

 

You can find this information in the syslog of the SIP Proxy machine, filter after mediaproxy dispatcher.

You can use an Asterisk server for this purpose and set MSP to route an ENUM number or a SIP alias to it. There is no conference server provided by the MSP, you must set it up yourself.

We advice a duration between 5 minutes and one hour. NAT traversal is performed by the server using Options packets sent every 60 seconds that trigger a response from the SIP UA, it is not necessary to send Register more often than 5 minutes for the purpose of keeping the NAT open.

A 3 GHz CPU is able to relay 200 calls at the same time. You may combine multiple media proxies to achieve load balancing and failover. MediaProxy 2.0 will provide more scalability on a per node basis.

At a minimum the following files and directories must be included in a daily backup procedure:

Please see http://openxcap.org for an up to date list. Currently the applications include pres-rules and resource lists.

Yes, it is possible to apply a fixed discount for calls routed to a E.164 number that had a destination found by an ENUM lookup. The discount is a percentage of the full PSTN price (from 0% to 100%) that can be set per ENUM top level domain where the lookup has been performed sucessfully.

Quota system provided by CDRTool is a statistical tool. It is not accurate down to the minute while it does provide on a platform basis a very low margin for the overal potential fraud. The quota check is performed every 5 minutes.

Use NGNPro related functions for Routes, Gateways and Gateway groups. A description of the LCR modus operandi is available online at OpenSER web site

 

CDRTool version 6.2 or higher provides a complete NGNPro client implementation written in PHP. The code is available in /var/www/CDRTool/provisioning/ directory. 

 

Some good practices for account creation: