CS1000E - Distance between Call Servers

[terrysdavid]

Hi,
I know that the CS1000E has a distributed architecture. Has anyone ever done a high availability installation where the Call Servers are in different locations. I know they can be in different buildings. But has anyone ever seen an installation where one call server is at one location and the other is at another location and the two sites are connected via MPLS?

 

[curtismo]

We have an HA install that we just upgraded from 6.0 to 7.5 where the two HA call servers are in different locations. The link between the locations is 8 blocks via a leased dark fiber network. The HA Ethernet subnet is a separate VLAN tagged within the 1 gbps pipe. Also have 3 cabinets at one site, 1 at another. Our RTT is usually 1 ms; looking at the last day the maximum RTT for this link is 2.5 ms.

Yes, it does work very well, and because I have local PRI at both locations, I can (and have) survived network and physical (power) failures. We also have sig servers at both locations, although we have had some issues having to do with IP routing with them, we are also protected here.

Now I am getting ready to add 6 cabinets plus a Geo Redundant CPPM at locations hanging off of a shared 6 mbps Metro Ethernet provider ring, with all six sites from 125 to 200 miles away. Our average RTT to these sites is 10-15 ms. This would probably be closer to what you are looking for in your MPLS network. I should have the install done by the end of August.

 

[Tman45]

Yes, we have a 6.0 HA system with two-cores at the main site and 1 core at the remote site (40+ miles away) that rides MPLS. We also have two SigServers at our main site and 1 at our remote location and our remote location is survivable.
We were one of the first Geo-Redundant installs back in early 2008. Consider us a guinea pig

We have had random card reboots and things of that nature that was initially resolved with patches. We have seen current random reboots (such as a MC32S) and Avaya is trying to figure that out right now. I think it has something to do with not enough ACL protection on the ELAN (so heartbeat related). We also had an issue where the remote Call Server was trying to steal clocking from the primary (also patch fix)

So, it should work just fine, you just have to make sure you are current on patching and chase down any strange issues

 

[nanningmiss]

We have a release 6 CS1000E.

3 Call Servers, one in Toronto, one in Winnipeg, and one in Vancouver. All connected by MPLS.

The ping time between Toronto and Vancouver is 60+ ms. Before the implementation, we had concerns about the distance (3-hour time zone difference). But the system has been up for about 3 years, and so far so good.

 

[wevertx]

The NTP does not give a distance limitation. It says that the round trip delay must be less than 30 msec and packet loss must be below .1 %. All the requirements and limitations are found in the System Redundancy Fundamentals NTP NN43001-507 in the Campus Redundancy section.

Wayne

 

[terrysdavid]

Thanks for all the feedback guys.