Avatar

In the first part we discussed how video services are evolving within enterprise networks. Content may be sourced from internal servers, BYOD end points or external content providers, thereby creating a mix of managed and unmanaged services. This has led not only to growth in traffic, but also a competition for actual resources between the different types of services.

We have discussed how these services are evolving, now moving to a per application, per session model which ensures that specific resources are allocated depending on the nature of the usage. Tools such as those provided by the medianet architecture, combined with changes in defaulting all traffic within the VPN session back to the corporate network, contribute to this evolution in session management.

Once again, we turn to Thomas Kernen to provide some insight into how recent technology improvements are designed to help with managing video traffic growth and enabling better content distribution models.

Q: What are the recent video evolutions in the video space that may impact network infrastructure and applications in the near future?

A: One of the major upcoming trends is the introduction of Ultra High Definition (UHD) systems, sometimes also referred to as “4k”. With a resolution of 3840×2160 pixels, this provides 4 times the resolution of High Definition. In specific market applications such as medical imaging, unified communications, broadcast TV, or surveillance, additional features may be introduced alongside the increase in resolution, enhancing services by including higher frame rates, high dynamic range, and a wider colour gamut.

 

Q: This will likely cause a significant increase in bandwidth usage, what tools are available to contain this?

A: That is absolutely correct. At source, before any form of compression, the raw output of a UHD signal is approx. 12Gbit/s. As one would expect, this is 4 times the rate of a 1920x1080p High Definition signal (approx. 3Gbit/s). Thankfully, the emergence of High Efficiency Video Coding (HEVC) will help to deal with containing the bandwidth requirements, not only for UHD but also for all resolutions. As a rule of thumb, HEVC, also known as H.265, MPEG-H part 2 or ISO/IEC 23008-2, more or less provides the same perceived quality for half the bitrate of the current codec generation Advanced Video Coding (AVC), which is also known as H.264/MPEG-4 part 10, or ISO/IEC 14496-10.

One must keep in mind that HEVC is still at an early stage. In its initial form it was approved in January 2013, hence implementations are still in their infancy, compared to AVC, which has been in existence for close to a decade. Over time HEVC compression efficiency will improve, as with previous codec generations, and applications will benefit from better encoding at lower bitrates.

 

Q: Will this have an impact on content distribution such as live events?

A: Whilst the compression efficiency will reduce the bitrate on a per session basis, video growth will nevertheless continue. To mitigate the bandwidth requirements for managed live events, such as town hall meetings, where a large audience is viewing the same content, IP multicast does provide the required scalability and minimal network infrastructure load.

Architectural improvements in IP multicast, such as Source Specific Multicast (SSM), greatly simplify deployments for such one-to-many architectures. While end-point receivers must support IGMPv3 in the OS and application stacks, there are solutions to work around those that don’t, thanks to SSM mapping.

 

Q: Isn’t IP multicast limited to wired connections only? What about BYOD devices connected over Wifi? Are they capable of receiving IP multicast streams?

A: Yes, both iOS and Android in recent versions support native multicast, and carry applications on their online stores that leverage this capability.

Combined with the introduction of the 802.11ac, the next generation Wifi standard which provides up to 3 times the bandwidth of 802.11n and Cisco’s Videostream technology, multicast delivery over Wifi is perfectly achievable.

Videostream provides a set of features to help manage the load, prioritize the activities on the network, and give a better quality of service. This translates into the following capabilities:

  • Stream Admission and Prioritization allows you to build classes of input multicast streams;
  • Radio Reservation Control provides enhanced capabilities to manage admission and policy controls. Admission and policy decisions are made based on the radio resource measurements, statistics measurement of the traffic, and system configurations; and,
  • Multicast-to-unicast allows a Wireless LAN Controller to snoop a wireless client application request to subscribe to an IP multicast stream, check a number of parameters, and if policies and bandwidth allow the stream to be transmitted, the access point will replicate the multicast frame and convert it to 802.11 unicast frames, delivering the video stream as unicast directly to the client.

These features provide the basis for reliable IP multicast delivery to wireless clients.

Emarin Video Blog 2

Q: So both wired and wireless can enjoy the benefits of native IP multicast delivery for scaling video services. What about remote sites that may not have IP multicast across the WAN, can they also benefit from a multicast transport?

A: Traditionally, in order to overcome infrastructure or service provider limitations whereby the WAN may not support natively IP multicast, the use of technologies such as GRE tunnels has been the norm. A new protocol simplifies all of this and not only connects the branch router, but can also directly connect endpoints that may not have native multicast capabilities: Automatic Multicast Tunneling (AMT).

AMT provides a multicast over unicast encapsulation mechanism which allows for on-demand tunnels to be set up between an AMT Relay connected to the native multicast infrastructure, and the AMT Gateway sitting across the unicast-only infrastructure, such as the WAN. The AMT Gateway initiates the tunnel request on behalf of an IP multicast receiver requesting to subscribe to a stream. The setup and teardown are dynamic and only last for the duration of the IP multicast session. An additional advantage with AMT is that the AMT Relay can use an anycast source IP address to help with load balancing and resiliency.

Q: So an end-to-end network infrastructure for live video delivery is actually possible across all types of infrastructures?

A: Yes. Native wired and wireless connectivity, combined with a tunneling technology for non-IP multicast enabled infrastructures, does provide the capability and flexibility to connect any IP multicast-capable device to the infrastructure, therefore ensuring that live video delivery can be achieved across the entire enterprise network.

 

References:

 



Authors

Eric Marin

CTO

Borderless Network Architecture, EMEAR