“Time is an illusion,” said Albert Einstein. But in a data center or any real-time application, time is all too real and must be managed very precisely. Consider a conversation on your smart phone while traveling 70 miles-per-hour down the Interstate. The call is handed off from one cell tower to the next. If the timing and frequency of the radios in each tower are not precisely synchronized, the call will drop.
Precise time synchronization is critical in every industry. Without it, financial services firms couldn’t track the stock trading or fraud attempts accurately, critical military missions could fail, critical processes would be unmanageable to orchestrate in manufacturing and healthcare operations, and self-driving cars couldn’t avoid accidents.
Enterprises are highly dependent on precise timing. If servers, storage systems, switches and routers aren’t properly synced, transaction logs would become questionable. For example, did a transaction take place on July 10 at 11:59:58 or July 11 at 12:00:00? The difference may not seem like much, but this can have a ripple effect across multiple interdependent systems that impacts SLAs and compliance. Timing failures can ruin video conferences with poor quality video; audio that isn’t synced with the video can cause jitter and cut out participants in mid-sentence. A lack of synchronization between servers can even impact cybersecurity, with inconsistent log files making it impossible to understand when and how an intrusion took place.
Deliver secure, reliable and precise time synchronization
Get a global, precise and secure Time-as-a-Service application for distributed enterprise systems through Equinix Fabric™.Read More
Precise time synchronization is critical in every industry.
How to achieve time synchronization
Every computing device has some form of clock inside. The crystals used inside these clocks are not expensive devices, but they drift randomly, that is, they lose time due to various causes, such as temperature, humidity and the age of the device. Unless the devices on a network are periodically resynced, eventually, they will all be using a different time.
Enterprise applications usually need to achieve 1 to 10 millisecond synchronization accuracy to avoid problems. Critical infrastructure, for the military or space program for example, may require 1-10 microseconds or sub-microsecond or even sub-nanosecond accuracy. To achieve these requirements, there must be a single, reliable source of time to which every device on a network can be synced. This has become especially difficult in hybrid cloud environments, where infrastructure is being deployed in private data centers and at the edge.
Today, there are three options for synchronizing time across distributed infrastructures:
The DIY approach – Using the global navigation satellite system (GNSS or GPS in the U.S.), companies can colocate a receiver and antenna in each data center and achieve sub-microsecond precision. In this model, companies retain ownership and control of the infrastructure, so they control their destiny. However, procuring, setting up, operating and securing the infrastructure is very expensive and time-consuming. It requires adding and training staff and is difficult to scale and support across geographies. Still, this is the most widely used approach today.
Internet time – Using ntp.org or other time service, companies can easily set up time synchronization for free over the public internet. However, this approach isn’t accurate enough for most use cases, and since there are no SLAs, it should be considered unreliable. Further, to use it, companies must open a port in their firewall, leaving their infrastructure susceptible to cyberattacks.
Time as a service – Legacy Time-as-a-Service (TaaS) offerings deliver microsecond-level accuracy while being easier to implement and manage than the DIY approach since the service provider deploys and maintains the infrastructure. However, these services are available only in a limited number of locations and are expensive to scale. In fact, they are quite expensive for the service provider because of the need to deploy specialized hardware at each location, which is why this approach has not caught on.
The best of all worlds – Software-enabled TaaS
As demonstrated by the multiple real-world case studies in the IDC report, Why Enterprises Need a Time as a Service for Digital Edge, sponsored by Equinix, “The benefits can accrue in operational efficiencies, service integrity, and business outcomes.”[i]
Equinix, for example, will announce general availability of its precision TaaS offering, Edge Precision Time, in Q1 of 2021. With Edge Precision Time, enterprises deployed on Platform Equinix® will be able to take advantage of direct, secure and reliable connectivity between source timing servers and distributed systems and devices via Equinix Fabric™. Leveraging Platform Equinix’s low latency and connectivity to more than 60 markets in 26 countries on 5 continents, Edge Precision Time will distribute reliable, predictable and secure end-to-end precise timing.
Using Edge Precision Time, for example, systems in New York, Chicago, Dallas and San Francisco could sync time using a master NTP server in New York. The direct and secure interconnection to this service via Equinix Fabric across Platform Equinix removes the need for any synchronization to take place over the public internet, which eliminates issues that could threaten accuracy, coordination, security and reliability.
Edge Precision Time can support both the Network Time Protocol (NTP), which offers millisecond accuracy, and the more accurate Precision Time Protocol (PTP), which delivers microsecond accuracy. The Equinix TaaS eliminates the complexity of the DIY approach, as well as the high upfront cost for hardware. The solution provides a comprehensive service as an OPEX rather than a CAPEX line item that includes all management and maintenance, as well as high availability. Like other Equinix services, Edge Precision Time can be set up with just a few mouse clicks in minutes. Edge Precision Time is currently available in North America, with plans to expand into APAC and EMEA at the end of 2020.
For more information on how Edge Precision Time can solve your timing synchronization challenges, download the Edge Precision Time data sheet.
You can also view our presentation to the ITSF.
You also may want to read:
Do Your Systems Have the Correct Time?
[i] IDC, “Addressing the Enterprise Need for Time as a Service,” Brad Casemore, My 2020, #US46256520.
A lack of synchronization between servers can even impact cybersecurity, with inconsistent log files making it impossible to understand when and how an intrusion took place.