Yesterday, Brian Heacox, Head of Product here at CFN, joined Viptela’s Lloyd Noronha, Head of Marketing to host the webinar: Three Steps to Reduce Costs & Cloud Proof Your WAN. The experts discussed the evolving costs & cloud performance challenges for the Enterprise WAN. You can check out a recording here.
During the webinar, a lot of great questions were asked and we wanted to take this opportunity to answer them in more detail:
Q: Is the monitoring of application latency integral part of your solution or are 3rd party products required to do this?
A: Viptela itself is able to measure the performance of the WAN from site to site. CFN has additional technology in beta mode. We plan to deploy telemetry nodes at all customer sites that constantly scan latency, jitter, packet loss, and b/w utilization for specific applications that users can define via our GUI, available for both on-prem, and cloud apps. Our product team would love to get your take on it, if you had an afternoon to spare.
Q: How do you address the latency when your overlay is over a third party underlay where you don’t have any control ?
A: While Viptela manages the overlay CFN operates as the performance tuned underlay. Our WAN platform is built to provide customers with fixed latency on the network from any site to any site. We do not oversubscribe the network so traffic is only rerouted in the event of an outage on a path in the core. In that case, customers are provided with a series of alternative latencies that they might expect. With CFN, Viptela is most crucial for managing the routing across multiple edge connections (multi-DIA, 1x EPL w/ 1x DIA, etc.) as the bring traffic to the closes of 60 CFN AppHUB data center nodes.
Q: How will WAN Optimization will fit into this SDWAN
A: There are a couple ways… WAN Optimization and SD-WAN generally have some overlapping feature sets when you compare strictly WAN Op and strictly SD-WAN solutions..our customers have taken different approaches. Some run it at the edge…but we see a number of clients running it within the core of the network within the CFN AppHUB. This way there is less of a hardware need as multiple sites use a single WAN Op instance to accelerate and control traffic.
At the moment we enable this by hosting client owned WAN-Op within our AppHUBs. However, we’ve got integrated SD-WAN and WAN Optimization on our road map so that it can operate as a full service.
Q: SIP needs reliable links both primary and secondary like MPLS how Internet link will fit into this?
A: Stability is crucial for applications like VoIP and Real-time video and MPLS and I agree that internet quality doesn’t always stand up to the MPLS level of performance…however when that happens, this is taking into account that the phone call may need to traverse 20 hops and 6 different IP transit networks in order to get from one caller to the next. At the peering exchanges where that traffic moves from network to network with BGP, the call traffic faces competition from the other public traffic being exchanged network to network. The result is jitter, packet loss, and service degradation.
The way to make internet at the edge a viable solution is to use it for its strength ($/Mbps) and not for its weakness (performance over long distance). When we deploy branches with internet edges, all traffic enter the CFN AppHUB network, a private network which runs multi 10Gbps MPLS in the core, via an IPsec VPN to a regional internet gateway at a CFN AppHUB. Thus, the traffic only rides the internet for the last mile where we see minimum traffic congestion and then rides the AppHUB low latency network to the destination.
More questions? Brian is ready and willing to answer! Send him a note here.