Experiences on Cisco CCM cluster over WAN? 2

[maraxman]

Hello,

does anybody have experience on Cisco CCM Cluster over WAN?. I am asking, because before we have bought CCM we were assured that they can do it. Now we have instalation of CCM cluster, but they do not want to support it! Technically there is only limitation of 40ms ping between servers. But practise is, that to apply so many patches on servers and aplications of CCM demands to be close.
Now we have official audit from Cisco and it says: get them to one place, because it is better. We had them only in different floors in the same building!!
One think is if it is works, second if it is supported and the last if it reasonable (because you have to patch system almost constantly)

vokoun

 

[BuckWeet]

Its not supported if its seperated by a WAN link, but if its in the same building, why isn't it on the same LAN? If it is on the same LAN (LAN being ethernet linked) why do you have such high latency?


BuckWeet

 

[maraxman]

Hello,
it was a LAN connection, but we had serious problem with pooling of datebases, and whole cluster crushed. Therefore they require to have ccm on the same switch. WAN cluster was our first dream! Never came true.

Bya

maraxman (formerly vokoun)

 

[BuckWeet]

I've seperated CCM servers onto different switches and it is supported. This was even going across layer 3 boxes. The buildings were seperated via fiber.

It is supported, because do they expect you to have your whole CCM cluster support 10K plus phones on a single switch. I think not..

Did you find out why there was such high latency?


BuckWeet

 

[maraxman]

Sorry for misunderstanding. I did not mean that we had 40 ms latency, I meant that this latency is the maximal latency recommended by Cisco to have a servers in cluster. It was only limitation to do cluster over WAN (of course to have wide band line)

maraxman

T-Mobile CZ

 

[BuckWeet]

Well, I'd get ahold of another Cisco Engineer, because I always seperate my servers onto different switches if possible.. I don't want a single switch failure to take out my cluster..

It is supported, if they try to tell you its not, tell them that "you expect me to put my whole redundant system at risk for this single switch"


Other than that I don't know what to tell you because I've done it, and Cisco has supported us..

BuckWeet

 

[scaine]

Buckweet,
Any ideas if you can have two boxes in a cluster on separate subnets? It's 100Mb LES circuit and we're looking to take our Publisher out of house to our hosting partner's site (latency is just off LAN speed - about 10ms). That way, the Publisher will be 10.97.x.x, while our two subscribers will be 10.96.x.x.

It's for Disaster Recovery - the idea is that if our Publisher is at a different location and we invoke to our DR site, we'll still have access to update our CCM environment. Ultimately, we'd look to run another subscriber out at the DR site, so that phone invocation is almost seemless.

Our Cisco partner are saying it can't be done, which is their way of saying that they won't support it.

Any thoughts?
Scaine.

 

[congogrey]

According to Cisco you need to have at least a DS3 between the servers to sperate them over a WAN link and keep them within the same cluster. Of course if you have a need to put IP phones at a remote location just use the central CM cluster to control the phones and use SRST at the site to keep the phones up in case of a WAN failure.

 

[BuckWeet]

Scaine, it can be done, I've done it, cisco supported it, and there was a DS3 linke between sites, but you say you have 100Mb so thats just like a normal lan connection there. Yes you can do them on seperate subnets as well..


BuckWeet

 

[maraxman]

another idea
Cisco cluster works only if active ccm totally fails (network fails). But this happened to me twice. ccm services works fine, but server for unknown reason was overloaded (server was uptime 22 days) Neither I could not connect from Terminal Services. This caused that you did not recieve dial tone, you could not dial etc. If ccm would be written well, why switching is not send to stand-by server? Cisco has cluster good for nothing.

bye

 

[Tomas74]

Dear Sirs,

I'd like to try CallManager 3.3
Could you prompt me where I can download it.
Please answer me to kolesnyk@ukrpost.net

Thanks.
Kolesnyk

 

[commsguy]

Scanine,

I have configured clusters over WAN links, works fine.

You should have no pronlems with the bandwidth you have.

Find another phone vendor.

commsguy

http://www.convetech.com

 

[talbotpat]

scaine - if latency is just off of LAN speed, AND you have the required bandwidth needed for SQL replication and intracluster signaling outlined in the Cisco docs, you should be good here. Sounds like the Cisco partner is either unaware of the Cisco support for WAN clustering or is simply afraid to endorse going down this road. Unless there is a reason or technical requirement you haven't addressed in your post that would adversely affect clustering over the WAN.

 

[bythespy]

You can run CCM on seperate subnets, I have done it in our lab.

You can run them across a WAN link, Cisco does not recommend that you spread a cluster across a WAN, unless it has sufficient bandwith to maintain the cluster integrity. The lowest circuit bandwith I have heard of was T1, but it was a small cluster, less than 1000 phones and a very static enviroment.

I have never read or heard a Cisco SE or anyone else from Cisco tell me that CCMs had to be on the same switch or subnet.


I have a 100MB fiber WAN and we have a pub/sub at our corporate location and we are deploying a second sub at our DR location. Our Cisco SE and Vendor have been involved from the begining of the design and we have had no pushback for Cisco or the vendor.







Mspivey
CCNP

 

[mtashiro]

It is supported
Does not have to be on the same subnet

Following are the requirements:

Every 10,000 busy hour call attempts (BHCA) in the cluster requires 900 kbps of bandwidth forIntra-Cluster Communication Signaling (ICCS). This is a minimum bandwidth requirement, and bandwidth is allocated in multiples of 900 kbps.

In addition to the real-time ICCS bandwidth, intra-cluster bandwidth is required for the SQL, CTI Manager, and LDAP traffic. The amount of additional bandwidth is dependant on the use of the system. For instance, the use of Extension Mobility increases the amount of SQL traffic between servers.

A maximum Round Trip Time (RTT) of 40 ms is allowed between any two servers in the Cisco CallManager cluster. This time equates to a 20 ms maximum one-way delay, or a transmission distance of approximately 1860 miles (3000 km) under ideal conditions.

 

[simonxts]

Does anyone know if the use of Cisco Attendant Console is supported over the WAN? There is a 512kb FrameRelay link inbetween.
Has anyone done this before? any tips or iedeas.

Thanx,
Simon

 

[AdmanOK]

We have all of our Attendants at locatations across the WAN, admittedly it is a 4MB ATM WAN so the speed is excellent. We are just about to put an attendant at a remote location that is connected over VPN so that should be interesting to see how it goes... I'll let you know if you are interested.

 

[simonxts]

Thanx AdmanOK for the response.

Simon

 

[dpaige]

We have 4 sites each with a sub connected via TLS and we also have a 45MB ATM backup. No problems since I got here.

 

[mtashiro]

This is suppotred and outlined in the Design guide. Here are the requirements:

• Cisco highly recommends that you replicate key services (TFTP, DNS, DHCP, LDAP, and IP Phone
Services), all media resources (conference bridges and music on hold), and gateways at each site
with IP phones to provide the highest level of resiliency. You could also extend this practice to
include a voice mail system at each site. Under a WAN failure condition, only sites without access
to the publisher database might loose a small amount of functionality.
• Every 10,000 busy hour call attempts (BHCA) in the cluster requires 900 kbps of bandwidth for
Intra-Cluster Communication Signaling (ICCS). This is a minimum bandwidth requirement, and
bandwidth is allocated in multiples of 900 kbps.
• In addition to the real-time ICCS bandwidth, intra-cluster bandwidth is required for the SQL, CTI
Manager, and LDAP traffic. The amount of additional bandwidth is dependant on the use of the
system. For instance, the use of Extension Mobility increases the amount of SQL traffic between
servers.
• Signaling or Control Plane traffic requires additional bandwidth when devices are registered across
the WAN with a remote Cisco CallManager server in the same cluster.
• A maximum Round Trip Time (RTT) of 40 ms is allowed between any two servers in the
Cisco CallManager cluster. This time equates to a 20 ms maximum one-way delay, or a transmission
distance of approximately 1860 miles (3000 km) under ideal conditions.
• The remote failover model requires Cisco CallManager Release 3.1 or later.

I have implemented this in several sites without issues.