02-09-2009 08:14 PM - edited 03-15-2019 04:06 PM
CallManager 4.1(3) at a central site, trying to simultaneously ring IPMA assistants which are in shared line mode at a remote site.
Because there are 10 lines at the remote office, CallManager causes a burst of SCCP messages, exceeding the reserved bandwith. As a result some messages get dropped or delivered out of order (due to TCP retransmission). This causes erratic behavior among the remote assistants' IP phones.
1) Is this a known problem with a fix?
2) What are the design guidelines about maximum number of âremote siteâ IPMA assistants that can be in shared line mode?
3) How much bandwidth (for SCCP signaling messages) must be allocated per IPMA assistant, in shared line mode, at a remote site ?
02-09-2009 11:53 PM
If you check SRND guide we have some calculations for SCCP traffic.
http://www.cisco.com/en/US/docs/voice_ip_comm/cucm/srnd/6x/netstruc.html#wp1045335')">http://www.cisco.com/en/US/docs/voice_ip_comm/cucm/srnd/6x/netstruc.html#wp1045335')">http://www.cisco.com/en/US/docs/voice_ip_comm/cucm/srnd/6x/netstruc.html#wp1045335')">http://www.cisco.com/en/US/docs/voice_ip_comm/cucm/srnd/6x/netstruc.html#wp1045335
Scroll down to the section titled:
Considerations for Shared Line Appearances
The main idea here is that 100 phones will each require about 4
packets (assuming SCCP, and 6.x). That is 400 packets generated quasi-
simultaneously by the UCM server (assuming they are all in the same
device pool). By default, your WAN interface will be able to accept 64
of those packets (coming in at 100, or 1000Mbps). The time it takes to
serialize 1 packet may be more than the time it takes CUCM to generate
400!
So the tail end of the 400 packet train will be dropped. Enevtually re-
transmitted, yes, but perhaps too late to avoid the perception of some
phones "lagging" in their response to UCM stimuli: they may ring late,
not ring at all if another phone picks up rapidly and all ringing is
cancelled, they may even re-register if the re-xmit sequences exceed
allowable limits.
Increasing the queue depth is the only weapon. Increasing BW is less
effective, unless you can approach the BW that connects CUCM servers
to the WAN router.
Few Shared line restrictions
1) Do not use shared lines for Cisco Unified Communications Manager
Attendant Console pilot points or hunt group members. The phone that
acts as the attendant console supports shared lines.
2) Do not use shared-line appearances on any Cisco Unified IP Phone that
requires autoanswer capability and do not turn on auto answer for a
shared-line appearance.
3) Do not configure shared-line appearances on the primary lines of the
phones; for example, if two phones have a shared-line appearance, only
one of the phones should have the primary line configured as shared (the
other phone should have the secondary line configured as shared).
Barge, cBarge, and Privacy work with shared lines only.
4) Cisco recommends that you do not configure shared lines for Cisco
Unified IP Phones, H.323 clients, and MGCP POTS phones; likewise, Cisco
recommends that you do not configure shared lines for H.323 clients and
MGCP POTS phones. If you configure the same shared-line appearance for a
H.323 client, a MGCP POTS phone, for example, NetMeeting, and a Cisco
Unified IP Phone, you cannot use the hold/resume feature on the H.323
client or MGCP POTS phone.
5) Cisco recommends that you do not configure shared lines for Cisco
Unified IP Phones 7905, 7912, 7940, and 7960 that are running SIP
because these phones cannot pick up held calls on shared lines nor can
they use the shared-line features Single Button Barge/cBarge, Barge,
cBarge, and Privacy.
6) Cisco Unified IP Phones 7906, 7911, 7941, 7961, 7970, and 7971 that
are running SIP have the capability of supporting multiple lines with
the same directory number in different partitions. However, configuring
and using other SIP phones with multiple lines with the same directory
number is not supported.
With info provided you should be able to answer your questions, its bot really IPMA dependant but shared line and BW issue.
The shared line may not be the best design for your remote sites, you may think in other strategy like Line group with non-broadcats algorithm.
You can contact PDI team which should be able to help you in the Design of the site.
HTH
02-22-2009 10:33 PM
Thanks for the detailed response. I agree that this is a shared line and limited bandwith issue..
1) What is âLine group with non-broadcast algorithmâ?
- Is this a method we can utilize on the CallManager 4.1?
- Where can I find documentation about it?
1) You mention "increasing the queue depth as ... weapon". How can we calculate the minimum amount of queue depth we need?
Thanks,
06-23-2009 07:52 PM
Hi,
I am experiencing a similar issue whereby i have a two remote sites on the same cluster (6.X) and a shared line amongst 10 phones. Of the 10 phones, one phone is located at a remote site and the other 9 are on the same site. The 9 phones ring simultaneously but the one at the other remote site will not. I have checked the CSS & PT etc its all correct.
In your post you listed the recommendations from Cisco regarding shared lines, is there a document i can find this information as i cannot see it in any of the above, thanks.
06-23-2009 08:12 PM
I found it..
Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community: