IP Phone SRV Queries for IP Phone Services

[goliver]

Hi,

It has been a while since I explored this topic (several years).  Our company writes IP phone services for UCM and with almost all companies using a DR site model from a cloud architecture perspective, I need to ask again (without labbing it all up and checking every phone model if someone already knows).

Are the IP phones doing SRV queries for IP phone services yet?  WAN capable Load Balancers are very expensive when they are not really needed if SRV records work properly.

Or I am also open to see how others are providing (non-shared IP based) clustering)for 3rd party IP Phone Services.  Even cisco's products use shared IP clustering in a lot of instances, but some are also SIP based, which makes it easy, but for xml based phone apps, there has really been a lacking area here.  Ideally, the UCM would allow multiple URLs and a configurable failover timeout in this area, but it does not.

Open for any answers regarding new phone firmware and also what others are using to accomplish it where a floating IP is not possible.

Thanks!

-Greg 

[Jonathan Schulenberg]

I have not encountered a single document that even implies SRV records are supported. While I cannot speak to the roadmap on the forum, I would be quite surprised to see this bubble up considering what else is in the hopper.

Load balancers are fun but keep in mind you don’t necessarily need a full inline load balancer (eg F5 LTM). All you really need is a the DNS A record resolution to be resolved to a known-operational host (eg F5 GTM or NetScaler GSLB). Note this would need to be a SAN in the Tomcat cert for TLS handshake to succeed.

As for cost, you may want to look at HAProxy: http://www.haproxy.org/
It’s on my wish list of things to lab/learn, which I never seem to have the luxury of getting to.

[goliver]

Yeah, there are plenty of free ones to accomplish it, but then you end up having to cluster those as well, etc :)

With the DR site topology it becomes a never ending *chase the network failure point" scenario unless the customer has layes-2 MPLS or DVPN or somethign between the sites. 

Most of our customers just do not have that type of network in place, and since the UCM handles it just fine, they sometimes expect everything else to work that way as well.

Thanks for your opinion and input - I was just hoping Cisco had something for us vendors in the pipeline (or already out :) ) for us partners in that area that I did not know about as of yet.  I'll keep chugging along with how we currently do things today (clustering at a single site with floating IPs) and allowing VSphere and Hyper-V, etc to handle it when the customers that do have more complex networks are in the mix. 

Thanks!

-Greg Oliver