IPTV Looks at IPTV
February 16, 2010
I have recently been evaluating video over IP technologies for Iowa Public Television. I mean if anyone should embrace IP technologies, who better than IPTV?
I've been gathering as much material as I can from various manufacturers to review and determine what devices or strategies best fit our application. I often find in the material information that actually modifies my thinking and shapes the direction. If I am trying to get video from one point to another, my first thought is to run a dedicated circuit and often, that is the best way to do it—but not the only way. So when someone suggests that the job can be accomplished by using the existing connectivity and network, I want to explore the possibility. Thus my research project and the plethora of data sheets, applications notes and white papers regarding video over IP that I am currently reviewing.
I reviewed one data sheet from Rivulet that mentioned the use of dynamic synchronization technology which, as with all things digital, was given an acronym, DST. The acronym stuck out for me because not only am I more familiar with its use meaning daylight savings time, but it was also used by Ampex as the acronym for their Data Storage Technology, a 3/4-inch tape format which also became the form factor for the Ampex Data Component Technology (DCT) tape format. Clearly we're running out of acronyms and we may need a standards committee to work on adding more letters to the alphabet to deal with the crisis. Think of it at the IPv4 to IPv6 upgrade for TLA's!
One of the simple truths that I have discovered in working with data networks is that once you have established how much bandwidth you require, construct the system with at least twice that much and budget for doubling that every two years for the foreseeable future. This is especially true when dealing with networks that will be carrying IP video.
The data sheet noted that when load planning for conventional IP networks, less than 30 percent of the capacity should be for real-time services, which is pretty standard based on what I have seen. So when I read that based on their DST performance, they could get to roughly 90 percent utilization, I wanted to know more. Unfortunately, I found out while speaking with their engineers that their HD video input option is not yet available and when it is, it will only be a 720p version and I am in a 1080i house. They didn't mention any plans for 1080p. This is still an intriguing product and probably worth further investigation.
The 90 percent utilization number really piqued my interest so I started talking to some networking experts I know from Cisco and asked how realistic this was. Of course no one wants to say it is impossible because as we all are told constantly, "digital makes everything possible." But the term that got bandied about the most was "managed expectations." Some see the concept of managed expectations as manipulative and seedy but the reality is that ex-pectations either need to be managed or they just happen and the end result can be very bad.
In the 90 percent utilization example we were discussing, the networking experts all agreed that 90 percent utilization was doable but that there would still be a performance impact and in this case, it would probably be on the non-real time services. How significant that would be depended on the non-real time service users. Typically those users are accustomed to brief delays in network availability and a minor slowdown for them while gaining a threefold in-crease in availability for real time may be a good cost benefit ratio.
However, if the slowdown on non-real-time is significant or if the non-real time service is mission critical, then the trade off may be less tolerable. Since in most broadcast facilities we deal with a hybrid of important real-time and non-real-time services, we have to look past just the numbers and under-stand the quantity and importance of the services being utilized. I think this is why in many cases I find myself leaning more toward independent standalone systems within my facility and only looking at common network connectivity-based systems for non-mission-critical services or when there is no other choice. The level of complexity, the difficulty in maintenance and the unforeseen interaction between disparate users of the common system need to be considered as part of the equation, not just the dollars for the equipment.
BROADCAST IS REAL TIME
If you think about it, it is one of the fundamental issues that need to be understood as we look at the conversion (or collision) of traditional video and IT infrastructures. Traditional video services have been real time since their inception. They have been transported over dedicated services both wired and wireless for 70-plus years and everyone that uses them has the expectation that they will perform as they always have. The significant changes in the performance over time has been an increase in the quality of the delivered content, whether it was black and white, color, or high definition. Because they were on dedicated circuits, the speed of the content delivered was constant and loading was not a factor.
IT services on the other hand have a much shorter history. They have been delivered over shared and public networks since their inception and the most important and noticeable improvements have been the speed at which they are delivered. Many of us remember the near ecstasy of getting a 28.8 modem to replace our 2400 baud service and suddenly, still graphics with 16 or even 64 colors could be sent from one computer to another. And now many users have computer connectivity at 8 Mbps or higher in their homes and 100 base T or Gigabit Ethernet at their desktops so the significant in-crease in performance for IT has been speed of delivery. Since IT is delivering 0's and 1's, the quality of content has remained relatively stable.
MOVING FORWARD, REGARDLESS
No wonder I'm in such a quandary when working in a hybrid digital facility. On the one hand I am pushing for more quality in the video and audio which increases the number of 0's and 1's needed. Simultaneously, I am increasing the number of users on the shared network that want faster access to this content. HDTV has barely started to catch on with consumers and we're already looking at the creation of content in 1080p and 3D. What will the bandwidth requirements for those be within a plant? At the same time, IT infrastructure is increasing the number of 0's and 1's that can be passed through per second. The increased capacity of IT is fairly predictable while the increase load for video is more tumultuous, increasing in dramatic explosions.
So planning a network that can handle this traffic is a challenge. And while there are solutions, most are complex and difficult to implement and man-age, especially at the station level where budget constraints often lead to understaffing and salary levels often have the brightest engineers looking else-where. It is no wonder that I tend to look at dedicated solutions to specific areas. By keeping them isolated and autonomous, I alleviate the opportunity for implementation issues in the new service from crippling the existing service while my staff and I try to figure out the issues, hopefully before the technology is obsolete and manufacturer support has ended.
I have heard the conversion to digital referred to as "the guaranteed engineer employment act." While that is humorous and potentially true, it could also be the "guaranteed unsustainable business model act." As we navigate through the digitization of our business world we must choose wisely, making sure that we are picking the right tools and technologies to do the job at hand.
Bill Hayes is director of engineering at Iowa Public Television. He can be reached via TV Technology.