Posted by John Bartlett, NetForecast | Aug 4, 2008

Connecting telepresence across the globe may require the services of more than one WAN service provider. I discussed this in detail in last week’s post, and talked about the problems that this presents. Today I want to take a look at one solution to this multi-network-with-QoS conundrum, and see how one company, IPV Gateways, solves it for their enterprise customers.

Here is the network diagram described last week. Each location has a telepresence suite. Two network service providers are used to connect enterprise locations in their respective geographies and interconnecting in New York. As described last week, New York becomes a bottleneck for bandwidth, is a stress point whenever the network is expanded, and adds complexity to managing QoS end-to-end for the telepresence application.

The folks at IPV Gateways solve this issue by creating a video connection point at their facilities in Toronto. The IPV connection point becomes a video quality peering point for multiple carriers so that traffic can flow from one carrier backbone to the second carrier backbone without losing quality. The new diagram looks like the one shown below.

Why does IPV not have the same bandwidth issues that the New York office had in the previous diagram? Like real-estate, it’s all about location. IPV has its equipment in the Telco Hotel in Toronto, where most of the world’s Internet service providers are also located. So getting a 100 Mbps or a 1 Gbps connection to the carrier backbone is as simple as running an Ethernet cable to the Meet-Me room in the tower. The cost of these in-building connections is much less than the cost of a DS3 running across town. IPV also has high-bandwidth connectivity in New York and in London.

There are a number of tasks to be performed in this video peering point to make the connections work. IPV first establishes the physical port connections to each carrier. These may already be in place, or may require opening new ports, depending on the policies of the carrier. Secondly, IPV connects to the enterprise MPLS VPN on each carrier. These VPNs will have different parameters for each carrier. Next IPV may have to translate the QoS markings as traffic passes from one carrier to the next. If Provider A is using DSCP AF41 to define the video class but Provider B is using CS4, then IPV translates that marking as packets pass through their equipment.

IPV will also provide security in the form of firewalls if the enterprise requests it. Note that in the scenario we have described here, the traffic is passing directly from one enterprise VPN to another, so no further security may be required. But depending on the policies of the enterprise, they may want the assurance of additional firewall support. IPV provides this support and can either manage the firewalls themselves, or make a management interface available to the enterprise security team to allow them to manage the security directly.

Because IPV is connected to the core of the network, they are often chosen as the right location for the MCU or bridge. To support this need, IPV leases out rack space within their facility so enterprises can install their bridge equipment and remotely manage it.

There is one more critical element, and that is the testing and monitoring equipment. IPV provides a monitoring tool they call ‘Insight’ that constantly tests the paths from the Video Peering Point to each telepresence endpoint location. A portal to this tool is available to the enterprise. This constant monitoring will quickly determine if a network problem exists, and can isolate that problem to the service provider or the LAN infrastructure of a particular site. The insight from ‘Insight’ saves hours of finger pointing and gets issues resolved quickly.

I’ll review some other vendors who are addressing this issue in the next few posts.