Alternate Channel Storage Backends Performance

Asterisk Asterisk Support
1

Hello everyone,

I have a system handling an average of 10,000 calls per day. I configured the channel storage backend to use cpp_map_name_id and have confirmed the configuration loaded successfully:

root@svr1:~# asterisk -x"core show settings" | grep Channel
  Channel storage backend:     cpp_map_name_id

I am using LibreNMS to monitor server resources and compared resource usage from two different days. Regarding memory, I observed minimal improvement. However, the CPU usage spiked significantly compared to the previous day.

My question is: what differences should I expect to see—and where—after enabling this channel storage backend?

Thank you

2

Which asterisk version are you using?

3

Hi George,

Asterisk 20.15.0

Thank you

4

Hmmm. 10,000 calls per day isn’t all that much, even if it was spread out only over 8 hours. I wouldn’t have expected to see any difference at all.

Can you provide some additional info?

  • Exactly how much did CPU utilization increase?
  • System resources available (CPUs, memory, etc)
  • OS, distribution, version
  • Install from packages or built from source?
  • If packages, from where?
  • If from source, was optimization enabled?
  • What’s the approximate calls-per-second?
  • Can you confirm that changing it back to ao2_legacy reduces the utilization?
5

I personally did not see any major improvements . :sad_but_relieved_face:

I really anticipated it should to be much more !

but did not see a increase in cpu vice versa a little decrease was there .

(from a overloaded node spiking to 30 load , going down to a consistent 22 load on a 20 core .)

or am i underappreciating the impact ?

stasis has a huge impact on the cpu . the average load is by 10 going up to 15

did not try stasis without cpp_map_name_id so i do not know if stasis update by itself has the same impact

that is my experience

6

The channels container, in comparison to stasis, is not used anywhere near as much.

7

Thank you for looking at this.

See inline responses.

Around 5%

16 GB Ram, 6 CPU Cores.

Debian 12.12.

root@svr1:~# uname -a
Linux svr1.domain.com 6.1.0-40-amd64 #1 SMP PREEMPT_DYNAMIC Debian 6.1.153-1 (2025-09-20) x86_64 GNU/Linux

From a custom package, yes, optimization was enabled.

Varies from 5-30

I did not try that yet.

8

I would think that 5% is well within a normal day-to-day variability.

Yeah, I would try flipping back and forth for a few days and see if there really is a statistical difference.

9

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.