Last week Dialogic announced the acquisition of NS-Tech - here is a snippet from the release:
“Now we can offer highly scalable and reliable software-based media processing, because the MRB can also control multiple media servers, which could be in different locations. PowerMedia XMS will also have the fail-over and reliability required in an IMS network. Integrating PowerMedia XMS with the NS-Technologies MRB enables PowerMedia XMS to have the performance and reliability of hardware-based media servers and MRFs, at a much more economical price.”
I'm highlighting the word "reliable/reliability" as it's used multiple times throughout the quote from Jim Machi, SVP Product Management and Marketing, but what does it mean to be reliable?
In my definition, to be highly reliable is the preparation for a catastrophic event, having a design for failure, how well things run when problems arise. Throughout my discussions with customers, I find that reliability is a subjective term based on the service being provided. Deploying a mission critical application will not have the same reliability requirements as say an IVR or voicemail system - not to say those service are not important rather they simply do not have the same reliability requirements. The reliability we're looking to feature here is on the high end of the spectrum with zero down time and very little interruption in service. For that level of reliability, the MRB is certainly the word.
A MRB (Media Resource Broker) is a standards compliant (RFC 6917 & 3GPP TS 23.218) software-based network element that sits in-line between the application server and the media servers. The MRB actively polls the media server cluster. If it detects a media server failure it will then move the calls from the failed media server over to the standby server.
So what exactly does this mean for reliability – and what does this mean from a user’s perspective?
The video below is me doing some testing in my office with the MRB. I built a one person conference through the XMS media server and the lower left is my local reflection video and the larger image is my conference coming out of the XMS media server. I am pulling the network plug from the primary media server and the MRB is detecting the failure and automatically (no application intervention) switches my conference to the secondary media server. There is no sound in the clip so you'll have to trust me the audio was restored in 100-200 ms. Video on the other hand was restored in 1-2 second loss in video with most of the time spent waiting for the new I-frame to come in. That is pretty reliable in my opinion.
How does this rate on your reliability requirements and could a MRB help solve some of your reliability concerns?
I look forward to hearing your feedback - and in my next post I'll take a deeper look into the use of the "scalability" and "performance" terms used in the same announcement.