Are Your Servers in Sync? Is Network Timing the Synchronized Swimming of the Digital Age?

As cloud-based services like online gaming and virtual reality (VR) continue to advance, latency has become a significant concern that needs to be addressed. The solution to this problem lies in utilizing the Precision Time Protocol (PTP). To provide context for the importance of utilizing PTP, we are dealing with timing accuracy in the range of hundreds of nanoseconds, which represent extremely tight timing requirements. This level of precision is crucial when operating distributed systems at scale, where various operations need to remain synchronous. An excellent example of such systems can be seen in the telecom industry’s 5G networks. Additionally, precise timing becomes even more critical when dealing with critical processes that operate infrastructure.

The legacy of timing and synchronization originates from the telecom industry, where it is vital to keep mobile base stations synchronized to ensure call mobility during handovers between base stations, or else calls could drop. In the past, telecom networks utilized Time Division Multiplexing (TDM) which reserved a timeslot for synchronization.

Here are some important facts about timing and synchronization in telecom networks:

• The radios found in Radio Access Networks (RAN) use oscillators to generate an accurate radio frequency. The oscillators must be synchronized with the base stations on the correct allocated frequency. The synchronization process is crucial to ensure that the mobile devices can seamlessly communicate with the base stations, and any deviation in frequency can cause signal interference and dropped calls.
• As we move into the 5G era, Time Division Duplexing (TDD) requires not only frequency synchronization of the oscillator, but also time of the day or phase synchronization of the oscillator. Ensuring accurate time synchronization is crucial for the precise transmission of information between the network and the user, as well as in the reverse direction, from the user to the network, at a predetermined time. To achieve that, all base stations and users on the network need to be automatically traceable to the same source of time. Thus, frequency and time of the day synchronization are both required.
• Both Global Positioning System (GPS) and Global Navigation Satellite System (GNSS) are widely used for various applications today. Originally, GPS was the only available constellation, but new constellations like GNSS have since been launched that can also provide accurate time information. GNSS is generally preferred for space-based communication systems, but there are concerns around the security of space-based time and frequency signals due to their low signal levels, making them vulnerable to jamming or spoofing. To mitigate such risks, one should incorporate network-based timing, although there is misconception that this approach relies on GNSS signals.  In reality, network-based timing combines signals from multiple clock sources distributed over a wide geographical area and brought together at a central point. This provides greater accuracy and resilience compared to relying on a single source. While interference with any of the sources could still result in some loss, other signals are available that could provide the necessary information. Additionally, if any of the sources were to be spoofed, they would look different from the other ones and will be rejected as such. So, in summary a combination of GNSS and network based timing architecture provides a strong foundation for frequency and phase synchronization which is critical for many modern applications.

Why is PTP Important for Enterprises and Data Centers?

There is a growing consensus that PTP will become the global standard for maintaining accurate time in computer networks, whether it is for data center applications or specific use cases like cloud-based online gaming, business video conferencing, and remote work. From today’s video conference calls to the new possibilities with remote work and collaboration, PTP has demonstrated significant benefits. Moreover, PTP is expected to be foundational for many products and services in the Hyperscaler networks, including the metaverse. 

Consider a data center application, where every server in every data center within a corporate network must know and agree on the accurate time for optimal performance. This level of synchronization is crucial, as all computers across the globe need to be precisely synchronized with universal time coordination (UTC). If computers can make decisions based on some common reference, it greatly improves the overall performance. Applications such as video conferencing and other cloud-based applications require precise timing between multiple servers and coordinating the time between them is critical for smooth operation. Coordinating time between multiple servers is akin to synchronized swimming in the Olympics, where all swimmers must perform their part of the routine at the same pace. If they perform at different paces, the routine will not look as it should.  Therefore, ensuring that all servers operate in sync is vital for efficient network performance, much like how synchronized swimming relies on all swimmers to perform their parts of the routine at the same pace.

In a distributed database application, achieving consensus is crucial. Typically, consensus is based on the alignment of all servers running the application. One way to achieve this alignment is by using Network Time Protocol (NTP) clocks. NTP is designed to synchronize time between devices in a network, but its precision is limited. For modern applications requiring top performance and low latency, NTP's precision is often inadequate. To address this issue, PTP was introduced in 2002. PTP uses a reference clock, such as the Grand Master Clock (GMC), to keep all devices synchronized to sub-microsecond accuracy. By constantly validating time across the network, PTP can achieve precision within nanoseconds, making it a more suitable choice for modern applications.

What is PTP? And What Makes it so Important?

In a nutshell, it all comes down to accuracy, or how close a devices measurement of time is to the actual time and precision - or how close different devices measurement of time are to each other.

In essence, achieving accurate and precise time measurements is crucial for a distributed system. PTP offers a solution to precisely synchronize clocks in such systems. A Stratum clock, which is connected to a network, is responsible for holding the current time and sending a time reference to any other device on the network that asks for it. This reference time is sent to other devices on the network via a data packet (known as the sync message), allowing them to update their clocks. Essentially, one device on the network acts as a timekeeper for all other devices, which synchronize with this central device to achieve accurate and precise time measurements.

Network latency (delay) can cause time inaccuracies when the data arrives at the receiving device. Several factors can contribute to network latency, including:

• The speed at which electrical or optical signals can travel across a wire or fiber connection.
• The variation in conversion time of transceivers, which can be influenced by factors such as temperature.
• The quality of the physical network components, such as routers, switches, and other network devices.
• The complexity and performance of the software stack that must be executed on the devices to send or receive network packets.

PTP deals with network latency by providing devices on a network with more accurate and synchronized timing information. The nanosecond level of precision helps to minimize the effects of network latency by reducing delays caused by differences in time measurements between devices.

Benefits of PTP

PTP has been utilized and endorsed by the telecommunications industry for years, and as 5G networks become more prevalent, PTP's extra precision and accuracy will become increasingly crucial in supporting new applications and services around the globe. In the near future, PTP is expected to become the standard for timekeeping in computer networks, serving as a foundational component for new applications and services that depend on low latency and reliable delivery.

PTP offers numerous benefits, not just for telco 5G networks but also for Enterprise LAN infrastructure particularly with the emerging needs in areas such as Artificial Intelligence (AI). PTP allows for the synchronization of GPUs across data centers, enabling significantly greater scale in AI capabilities that are difficult to achieve with current CPUs alone. The precision of PTP is essential for guaranteeing the synchronization of not just modern-day computer networks but also advanced systems that will be used in the networks of the future.