GV Blog

IP Video: What We've Learned So Far

m☰nu
IP Video: What We've Learned So Far



Translation(s) Available: English | Portuguese / Português | Spanish / Español


The last few years have been an exciting time for the broadcast industry. We've seen the rapid development and deployment of IP and COTS-based infrastructures introduce new workflows to streamline operations and reshape how we build facilities.

With all the options and flexibility that IP-based infrastructures enable, additional complexity is also added to the system. On top of existing broadcast standards for transporting video over coax, such as SMPTE 292M, a plethora of emerging standards have been introduced into the marketplace that, although functionally solved a problem, generated large interoperability issues that inhibited growth and mass adoption.

With conflicting methods of transporting video over IP (NMI, SMPTE 2022-6, ASPEN and others), customers were hesitant to deploy IP solutions that were proprietary to a single vendor. Who wants to risk deploying a solution that might quickly be superseded by a more flexible emerging standard? With the work of industry associations like AIMS (Alliance for IP Media Solutions), we have seen the industry coalesce quickly behind SMPTE 2022-6 and the proposed SMPTE 2110 standard. Having a functional set of standards with which a broadcast or post production facility can deploy has allowed for the rapid adoption of IP infrastructures into workflows of all sizes. We have learned that with new technology comes risk, but with multiple vendors working toward a common goal, under a common set of standards, that risk can be mitigated.

The speed of technological innovation is one of the greatest strengths of migrating to a COTS infrastructure, but it also makes for one of the greatest challenges. is a generalized chart of the advancement of baseband data rates, compared to the advancement of IP switch data rates over time. When you compare that graph to , you see that COTS manufacturers have been pushing data rates higher and higher, to a point that is well beyond the requirements of the broadcast industry.

Just three years ago, 10GBbase and 40G aggregate data rates were the only cost-effective solution for transporting video. 10G was great for transporting 1080P and lower resolutions, but 4K UHD with its 12G data rate was challenging. The signal either had to be split across multiple physical interfaces or a light compression such as TICO (4:1) was required. With the emergence of 25G base data rate interfaces and 100G aggregate interfaces, an elegant path to transporting 4K UHD via IP is now available.

Using 25G on edge devices has led to some additional challenges in terms of network efficiency. 25G ports carry a significant cost premium compared to 10G ports. If a device, such as a camera running 1080p, uses a 25G port, you pay for a 25G port but only use a fraction of its capabilities. On one device that might not be a big issue, but with 20 cameras (or more for some facilities) the inefficiency of the design is multiplied significantly. If you scale that out to all devices on a system that only puts out one or two 3G signals, you have a large waste of CAPEX funds, allocated bandwidth, and physical switch space. Flow aggregators, such as Grass Valley's GV Node, have been playing a larger part in delivering cost effective solutions that allow for aggregation of lower bandwidth signals onto larger aggregate data interfaces. For example, GV Node can take up to 144 3G bidirectional signals, which could easily be 72 10G connections taking up multiple network switches, and aggregate them into 12 40G connections, significantly lowering the physical layer IO count, and maximizing the bandwidth used on those connections. See .

With the physical layer issues considered, and a set of standards to deploy against, the issue of COTS switch compatibility was a hurdle that had to be addressed. Not all switches are the same.

Just because a switch manufacturer claims that it can transport 10G or 25G, does not mean it can handle full bandwidth, across all ports, all the time. Most switches are designed to transport data in a "best effort" manner, meaning that they will make their best effort to get the data to the destination on time. If it doesn't get there, the switch will keep trying. That's great for file-based workflows and using protocols that are designed to use high latency or buffered connections, but it's a disaster for real time video production.

With real time video, we cannot simply accept that the data will be transmitted at a later time. The solution has to be designed so all source flows can switch to all destinations, at any time, without delay or buffering. Most switch manufacturers now have switches designed to do exactly that, but require special firmware, licenses, ASICs on the switches and control software. The first customers to deploy large scale COTS infrastructures bore the risk and learning curve that was required to find a functional solution. Now that we have many large scale deployments, the risk has been largely mitigated.

Today's benefit in using COTS switches is simple: Multiple COTS switch manufacturers have deployed turnkey switch solutions that do not require the end user to know what firmware, licenses and ASIC's are needed to meet customer requirements. The fully vetted COTS solution drastically simplifies the project's design and installation phases. That makes life in an IP world that much easier.

- Robert


Further Reading:

 Comments