Protecting Broadcast Quality! Basic Knowledge of PTP Network Design
Hello everyone! Today, I will talk about PTP network design, which is important for those in the broadcasting industry. Based on a presentation at IBC 2019, I will introduce common pitfalls and how to avoid them.
Part 1: Pitfalls and Points of Caution in PTP Network Design
From the presentation by Thomas Kernan of Mellanox, I have picked out points of caution for PTP network design in SMPTE ST 2110 systems.
The Complexity of PTP
PTP sounds difficult just by hearing the name, doesn't it? But it's not as complex as designing the International Space Station (laughs). The important thing is to create a blueprint that fits your system. It's just like creating an original recipe!
Key Elements of PTP
The essence of PTP is "time" and "accuracy." Imagine the entire system moving in sync, like breathing together. Accuracy is truly the heart of the system. If this drifts, the entire rhythm will fall apart.
PTP Basics
PTP is based on an established standard called IEEE 1588. Think of SMPTE ST 2059-2 as a profile customized for SMPTE ST 2110 systems.
The maximum offset between the slave and the grandmaster is usually 1 microsecond. In reality, it is around ±500 nanoseconds. That is an unimaginably short amount of time!
Endpoints must be designed to operate properly not only during daily use but also in unexpected situations.
PTP stacks are implemented differently by each vendor. Stability, filters, and control loops are key points.
Network Design
Network design is like moving the hands of a clock accurately. It is important to convey timing precisely and ensure the necessary accuracy.
It is common to use PTP-capable devices. These are like excellent interpreters that can speak a special language for PTP.
Whether to use a boundary clock or a transparent clock depends on the situation. You choose based on the scale and structure of the network and the number of endpoints.
Project Management
Project management holds the key to the success of PTP network design. Let's list some common pitfalls:
Sometimes quality is sacrificed because of being too rushed by deadlines. Let's set clear quality metrics at each stage.
It is important to understand constraints and find workarounds. Instead of saying "we can't," let's think about "how we can do it."
Consider the impact of other emergency projects as well. It is important to look at the overall balance, not just one project.
Timing Strategy
Timing strategy is the heart of a PTP network. Here are some common pitfalls to watch out for:
Be careful when changing default values. Thorough testing is required.
Be cautious when setting the Priority 1 value. It must be handled carefully because it will continue to be used even if a problem occurs with the primary reference source.
The placement of the Grandmaster should not be a major issue if the system is properly designed. However, it is better to be safe and double-check.
It is important to maintain consistency in settings across the entire PTP domain. Just like an orchestra, it is essential that the whole is in harmony.
Network Design
Network design is the backbone of a PTP system. Here are some points to help you avoid pitfalls:
Using PTP-capable devices can reduce CPU load and jitter. This is like hiring professional athletes.
Boundary clocks are useful for controlling PTP traffic. Think of it as directing traffic.
IPv6 has several advantages, such as the use of link-local addresses. It can be considered a forward-looking choice.
Multicast and mixed mode are generally preferred messaging methods. They allow for efficient information transmission.
PTP traffic can be sent on a separate channel from primary data. This is like creating a dedicated lane on a congested road.
You can isolate PTP traffic using VLANs, LAGs, and VRFs. This makes it less susceptible to the influence of other traffic.
Security
Security is like armor that protects a PTP network. Pay attention to the following points:
Authentication, authorization, and accounting are required for connections to all devices. This is like a security check when entering an important building.
Use encrypted communications such as SSH and HTTPS. This is like using a special safe when transporting sensitive information.
Protect physical ports as well. You need to secure all access points, including console ports, serial ports, auxiliary ports, and front panels.
Disable unused services. If there is an open window, there is a risk of intrusion through it.
Perform threat modeling for both hosts and the network. This means considering defensive measures from every angle.
Monitoring
Monitoring is like a regular health checkup to check the health status of a PTP network. Focus on the following points:
Use appropriate monitoring tools to understand the state of your PTP infrastructure. This is like a fitness tracker that constantly checks your body's condition.
You can capture messages offline or online to diagnose problems. This is like an examination to investigate physical ailments in detail.
You can set alarms to sound when a specific offset is reached. This is like a mechanism that issues a warning when body temperature exceeds a certain level.
You can use 1 PPS output to compare signals between the reference source and the endpoint. This is like comparing your own watch with an accurate clock.
Diversity and Failure Testing
Diversity and failure testing are like training to increase the durability of a PTP network. Focus on the following points:
PTP redundancy is not a standard feature. You must design it yourself.
Source diversity, antenna diversity, and frequency and time traceability are essential for a robust PTP infrastructure. This is like getting nutrition from multiple sources.
You can create a list of trusted clock IDs using an acceptable master table. This is like making a list of trusted friends.
Test partial and total failures to evaluate system resilience. This is like training under various conditions.
Measure recovery time and identify bottlenecks. This is like finding and improving performance weaknesses.
The Gap with Reality
Bridging the gap between ideals and reality is an important part of PTP network design. Focus on the following points:
Communicate directly with vendors rather than just relying on data sheets. This is like asking the author questions directly instead of just reading the book.
Investigate existing projects to learn best practices. Borrowing the wisdom of predecessors is a wise approach.
Every project is unique. One size does not fit all.
Industry publications and reference designs are valuable sources of information. They are like advice from experienced mentors.
PTP is still a relatively new technology. Since experience may be limited, it is important to maintain a mindset of continuous learning.
Key Points
Finally, here is a summary of the key points to keep in mind for PTP network design:
Make a plan and stick to it. However, don't forget to be flexible when necessary.
Verify your design and find compromises where needed. Seeking perfection can sometimes disrupt the overall balance.
Test the entire system end-to-end. Partial testing might cause you to miss potential issues.
Test failure scenarios to evaluate system resilience. Preparing for the worst-case scenario ensures you won't panic when facing actual problems.
Work with vendors to resolve issues and enhance features. Good relationships lead to long-term success.
Part 2: Practical PTP Experience and Tips
From a presentation by Greg Shea of Telos Alliance, here are some practical PTP experiences and lesser-known options.
PTP without a GPS Reference
Surprisingly, AES67 and PTP can function without a GPS reference. This is called an 'arbitrary clock.' In a closed LAN or a self-contained facility, it is sufficient if all devices reference the same clock. However, there are a few things to note:
The open-source PTP4L client does not work with times before January 2, 1970. You can't use a time machine (laughs).
Using NTP to provide time to a PTP master is not recommended due to issues with accuracy and stability.
While it is technically possible to transmit PTP over a WAN, it is better to avoid it because jitter will increase.
The Necessity of PTP-Aware Switches
PTP-aware switches improve synchronization accuracy by correcting the timestamps of clock packets. This acts like fine-tuning the hands of a clock. If you do not use a PTP-aware switch, maximum QoS priority inversion delay may occur. This delay varies depending on packet size, wire speed, and the number of switches. PTP-aware switches recognize this delay and correct the timestamps, enabling more accurate time synchronization.
Peer Delay Request/Response Mechanism
The peer delay request/response mechanism is used to compensate for network delay with the grandmaster. Without it, a delay of several tens of microseconds usually occurs. Interestingly, RTP timestamps are not affected, so phase alignment of multiple streams is possible. The delay manifests as a playback time error in analog output, but a delay of several tens of microseconds corresponds to a phase error of only a few samples, which is inaudible in radio audio. However, using peer delay request/response can reduce clock stability and degrade audio quality. This is because time noise is generated in both the delay request and the initial transmission, increasing overall noise. It's like trying to make extra adjustments and ending up with lower accuracy instead.
Statistical Minimum Filter
Here, I would like to introduce a little trick. Using a statistical minimum filter can effectively improve PTP synchronization recovery. This filter selects the synchronization time with the least delay from a series of synchronization times. By doing this, you can obtain the most accurate timestamp without being affected by jitter or noise. It is like choosing the most accurate clock from a large collection of clocks.
The Importance of Internal Oscillators
When implementing PTP recovery, it is important to use a stable oscillator. This can be considered the heart of the entire system. General crystal oscillators have large tolerances and are susceptible to temperature changes, which can reduce clock stability. This is like an old clock that expands and contracts depending on the temperature. On the other hand, temperature-compensated oscillators have smaller tolerances and are less affected by temperature changes, which helps improve clock stability. This is like a high-end watch that keeps accurate time in any environment.
Conclusion
So, we have looked in detail at the pitfalls of PTP network design and how to avoid them. Certainly, PTP is a complex technology with its own unique challenges. However, with proper planning, testing, and cooperation with vendors, you can build a robust and highly reliable PTP infrastructure.
Having worked in this field myself, I feel that the evolution of technology is truly rapid. PTP technology is also continuing to evolve every day, and in the future, it will provide even higher precision and stability. I am sure you are all looking forward to the evolution of technology from here on out. We are now standing at the entrance to a new era of broadcast technology. It is truly exciting to think about what kind of innovations will happen next.
Technology is certainly important, but the wisdom and experience of the people who master it are just as important. I have felt that it is essential to keep learning, catch up on new information, and above all, maintain an attitude of taking on challenges without fear of failure.