The growth of the digital economy and distributed computing has put an increased focus on time accuracy in networks. As digital transaction volumes increase, manufacturing operations become more sophisticated, and IoT functions rely more on real-time data, time synchronization across distributed workloads and databases has never been more important. Operating without accurate time can result in transactions processed out of order across multiple servers, expired data overwritten on current data in distributed databases, or a glitch in the delivery of live streaming media.
Download the Equinix Precision Time™ Data Sheet
Get more information and technical specifications for Equinix Precision Time, a secure PTP and NTP Time as a Service for distributed enterprise applications on Platform Equinix®.Download Now
Time protocols meet different time accuracy requirements
Network Time Protocol (NTP) and Precision Time Protocol (PTP) are the two primary methods of network time synchronization. Some applications simply require all nodes to be in sync with each other, with no outside reference. But others must use legally recognized time sources like Coordinated Universal Time (UTC) and be able to validate how far off their internal time is from the legal standard.
NTP is designed for syncing time over the internet or a multiprotocol label switching (MPLS) network. NTP can usually achieve time accuracy in the single-digit millisecond (ms) range, even over a wide area network like the Internet, which is sufficient for many commercial and industrial applications.
PTP was designed to achieve more precise synchronization on smaller networks. Its initial use was in industrial automation to control robots on closed networks. PTP can take advantage of hardware timestamping capabilities in existing network cards to achieve better results. Some network switches and routers also offer PTP-specific features like Boundary or Transparent Clocks to help compensate for their propagation delays.
Both protocols are doing the same thing—synchronizing computers by exchanging timestamps. NTP was designed with the Internet in mind. Its algorithms can make the best time possible out of sources coming from networks outside your administrative control. PTP is capable of better performance when a network is designed with the proper hardware.
Time accuracy requirements vary for specific business operations or industries
Synchronization accuracy to a common reference is critical to track information as it moves through different nodes across enterprise networks. The level of accuracy required can vary based on the needs of specific industries and can be met with close attention to network engineering for time transfer. For this article, I’ve created three groups of time accuracy levels: 1s of milliseconds, 10s of microseconds and 100s of nanoseconds. The groups are based on how much time a business needs to react or respond to a certain volume of events and how much resolution is required from one event to another.
- 1s of milliseconds (ms): This level of time accuracy is adequate for a large variety of applications, including log file creation and maintenance required for audits, logging and security use cases, reconciling stock trades and detecting fraud in financial services, and transaction processing for small and midsize retail businesses. The NTP time protocol can typically provide this accuracy level running on any device, and there are many options for pointing to available NTP servers on the Internet. However, these servers may not come with any guarantees about the quality of their time.
- 10s of microseconds (μs): Time accuracy at this level is necessary for power substations to transfer power smoothly and maintain power supply integrity, sensor networks to collect data accurately in manufacturing, patient monitoring and in operating rooms in healthcare, and high-volume transaction processing for large credit card companies and online retailers. At this level, obtaining time from the public internet will not be sufficient; network operators will need to have a design for how to get an accurate time base into their network.
- 100s of nanoseconds (ns): This level of timing is crucial for high-frequency traders, consumer device pairing and tower-to-tower voice synchronization in telecommunications, and broadcast networking of live and streamed events captured with multiple cameras and audio recording devices. Data and frames from sensors on self-driving cars must be processed and organized in the exact order of collection to accurately assess what even the human eye cannot see. Also, military defense operations that sync drones to GPS require precise time synchronization. To meet this level of accuracy, the time source for the network must be as precise as possible, and network elements must be chosen with PTP hardware support in mind.
Assess challenges with deploying and maintaining timing infrastructure
When businesses need to distribute and manage time across regions, countries or continents, maintaining a high degree of precision becomes complex and expensive. In a white paper, IDC observed that:
“…enterprises frequently find that it is difficult and expensive to manage and distribute time over great distances while maintaining a high degree of precision. Although NTP and PTP methods can adjust for latency, enterprises also require a network that can deliver minimal jitter.”
Whether time synchronization is being managed across the public internet or via smaller networks, there will always be inaccuracies to manage. With the public internet, traffic spikes can delay the transfer of time from the source to the client. When using network connectivity, the path between the client and server must be as symmetrical as possible in order to ensure time transfer accuracy. Asymmetry occurs when timing packets are delayed in one direction, resulting in an offset at the client that will be undetectable without access to another time reference.
Operating without accurate time can result in transactions processed out of order across multiple servers, expired data overwritten on current data in distributed databases, or a glitch in the delivery of live streaming media."
In addition, each network switch may add jitter to the time transfer. Introducing too much jitter will affect the quality of synchronization. Routers that store and forward packets can generate asymmetry, especially under high traffic load. Network switches that have PTP Boundary Clock or Transparent Clock support can compensate for this additional jitter. But even with those present on the network, all it takes is a single misbehaving switch to undo the benefit of all that additional hardware.
Distance also matters, because longer connections need more hops, introducing more opportunities for jitter and asymmetry. While it is possible for a Stratum 1 NTP server to support a worldwide network, the accuracy will be lower if the time packets must traverse multiple connections between metro areas. It’s one thing to sync time from one rack to another in the same data center; doing so from across the world requires a better solution. That’s when it might be time for IT organizations to consider using private interconnection to keep precision time.
Deploy Time as a Service infrastructure to simplify access to accurate time
Time infrastructure in the form of Time as a Service can address the limitations of using NTP, PTP and Stratum servers over the public internet or a MPLS network. Equinix Precision Time™ on Platform Equinix® provides reliable NTP and PTP infrastructure, makes it easy to scale, and shifts costs from CAPEX to OPEX.
Whether time synchronization is being managed across the public internet or via smaller networks, there will always be inaccuracies to manage."
Equinix uses redundant and strategically located GPS antennas, receivers, grandmaster clocks, and the high-performance network backbone of Equinix Fabric™ to deliver secure, reliable and precise time synchronization. We’ve established Time Masters in key metros worldwide with a GPS antenna on the roof of two Equinix International Business Exchange™ (IBX®) data center buildings in each metro. Time Masters are currently located in Europe, North America, and Asia (two metros each) and we are in the process of adding more. Time accuracy can be accessed from other metros in your region using Equinix Fabric for private, software-defined interconnection. Companies can access one or both metros, allowing them to achieve redundancy while eliminating the high costs of maintaining their own GPS antennas.
Equinix uses extensive monitoring and reporting tools to ensure that the precision timing infrastructure is meeting the Service Level Agreement for synchronization accuracy and availability. In addition, Equinix Precision Time has integrated industry-leading hardware and software to constantly monitor GNSS signals, mitigate against the threat of jamming and spoofing and take corrective action.
Finally, I tell customers that getting started with Equinix Precision Time is as easy as ordering pizza. If they are already using Equinix Fabric, Timing as a Service infrastructure can be provisioned in minutes.
To learn more about how your business can use Timing as a Service to simplify network time synchronization, read the Equinix Precision Time data sheet.
You may also be interested to:
View this Equinix Precision Time video
 IDC white paper, sponsored by Equinix, “Addressing the Enterprise Need for Time as a Service,” Doc #US476646121, May 2021.
Time infrastructure in the form of Time as a Service can address the limitations of using NTP, PTP and Stratum servers over the public internet or a MPLS network."