I need help please , i have Cisco IP telephony cluster which contains " 2X CUCM "Publisher and Subscriber" ,2 unity connection , 2 Presence and 2 Contact center express and one voice gateway". My issue that the ISP can only extend PSTN lines to the DR , which means the VG should be install on the DR . My questions
1- what is the best design which guarantee HA and consume lowest bandwidth between Main and DR for signalling and RTP?.
2-I need also to know if the subscriber is responsible for TFTP and all my phones registered to subscriber , so i have to put subscriber on main DC on the same location of my end users for internal calls and internal signalling or i have to put it on the DR behind the voice gateway for the signalling between subscriber and VG for outgoing and incoming calls?.
Notes : All end users will be on the MAIN DC building.No.of users around 250 users .
The region setting between IP phones and gateway should be g729 if you are looking to preserve WAN bandwidth.
Since you have all the users at Main site there is no need to have a server moved to DR site.
You may check the Centralized Services section of the following link for more details
- Do rate helpful posts -
Thanks a lot.
The idea to send some servers to DR ,for Redundancy because if any issue happened on Main the DR will serve my users .
One question here:-
1- As i understood from yours that signalling will be from users to the VG , not from CUCM to VG regarding incoming and outgoing calls ?.
Signaling will be between cucm and gateway / endpoints , however RTP will flow directly between endpoints ( ip phones and gateway ). RTP does not flow through cucm, unless you have a software MTP involved in the call that is registered to cucm.
Having a server at remote site would be a good option as it will also allow you to add dome extra users on that site in the future and also you can look at the possibility of configuring SRST on that voice gateway on DR site to take care of the phones in case the subscriber is down.
- Do rate helpful posts -
SRND provides perfect explanation on the requirements for clustering over WAN. I can share some of the bullet points here:
1. You need to make sure that the RTT latency shouldn't cross 80 msec.
2. You need to have at least 1.5 Mbps bandwidth for ICCS (intra-cluster communication) which can accommodate up to 10000 BHCA
3. You need to have at least 1.5 Mbps bandwidth for DB replication
4. In case you enable CTI manager and/or JTAPI, you need to accommodate their bandwidth.
5. Each remote phone takes 20kbps for signaling
6. RTP stream bandwidth depends on the used codec.
PS: These numbers are from Cisco SRND which are the ideal numbers. However, for low number of endpoints, you can have lower specs. I managed clustering over WAN for 1000 phones using 2 Mbps dedicated between UCM nodes and worked very well.