The Five Elements – Part 2: A Timekeeping Architecture that Fits

Acquiring UTC from GPS requires taking the signal off the air and delivering it to the clients (PCs, workstations, servers, controllers, etc.) that rely on accurate time for event synchronization and time stamping. Since very few clients come equipped with reliable network time synchronization software, this implies that besides the GPS receiver itself there has to be a way to distribute time from the GPS receiver to the clients. In other words, the organization needs a time distribution network—ideally an architecture of deployed time servers and time clients.

Why the need for such an architecture? Because time servers acquire time from GPS receivers and distribute time in response to client requests. Depending on the size and topology of your network, you may want to install multiple time servers in a certain configuration:

• To support multiple LANs
• As a backup
• To handle peak load volumes of client requests
• To handle special time-sensitive applications

As noted earlier, distributing time via a WAN introduces delays that can be avoided when each LAN takes UTC from GPS. But there are also good reasons to have multiple time servers at each LAN—as a backup, for example, in case one of the time servers goes down or becomes overloaded due to a spike in requests. In these scenarios, clients may have a primary and a secondary (or even tertiary) time server to reference in case of a problem. Also, one time server might itself be set up to receive time from another time server (say, if its connection to GPS were somehow lost). The purpose of these types of architectures is to make the network self-healing with respect to timekeeping.

The choice of GPS receiver, the choice of network time servers, and the architecture of the network are key to delivering an accurate time to the network. Below are a few factors to consider when making those choices:

Server performance — While the ability to synchronize hundreds of thousands of clients is catchy marketing, the real test is the volume of peak load requests the server can handle simultaneously—while maintaining accuracy and availability.

10 / 100 / 1000 Base-T Ethernet — A time server that supports up to gigabit Ethernet will accommodate networks of today and the future.

Redundant time sources — A time server may employ various strategies to ensure continuous UTC availability—such as multiple UTC Time Sources or a GPS receiver with NIST dialup as a backup.

Single satellite timing — In urban canyon environments where satellite visibility can be limited or where roof access is restricted, an automatic single satellite timing mode provides accurate time with intermittent satellite coverage. It can also be used to track satellites using a window-mounted antenna. Built-in time reference—When access to UTC is interrupted, a network should be able to maintain the required timing accuracy for a period sufficient to enable continuity of the supported applications and business mission. For example, Rubidium atomic time references can be employed in those time servers that acquire time from GPS and which redistribute the time to other time servers or to clients.

Once the network hierarchy is in place, the next issue is how to manage it. As previously noted, time is not something most organizations want to think about—so management of network time distribution should be simple and straightforward. Management occurs on three levels: 1) monitoring and controlling the network devices (e.g., setup and configuration); 2) comprehensive management of clients and devices as a network; and 3) providing a verifiable audit trail of the time synchronization across that network.

Microsemi provides synchronization services that assist customers with the planning, deployment and maintenance of synchronization infrastructure. Services are designed to lower costs, streamline processes, ensure quality, and deliver the highest level of performance from your synchronization network. Visit Timing & Synchronization Systems and learn how we can help provide you with comprehensive solutions across a wide range of applications.

Read the previous articles in this series, here, including “5 Essential Elements of Network Time Synchronization,” “The Importance of Network of Time Synchronization,” and “Part 1 of 5 Essential Elements.”

Please connect with me on LinkedIn and also follow Microsemi on LinkedIn.

Tags: Coordinated Universal Time, Network Operations, Synchronization, Timesharing, Timing, UTC