中文
We previously explored why this has been so difficult for datacenter operators, especially in comparison to previous generation architectures. But our latest work with customers has demonstrated a path forward. Importantly, at the core of any successful accuracy and performance timing strategy will be the ability to verify synchronization in the live network.
This calls for a dedicated tool for measuring datacenter timing and synchronization with microsecond accuracy.
Measurement techniques for datacenter synchronization
The endless glut of traffic generated by video streaming and other applications has driven rapid growth in datacenters, even before the arrival of anticipated 5G services. Datacenter servers are often physically separated across remote locations in less-than-ideal conditions, threatening performance.
Sure, manual methods can be used to determine timing offsets but there are several downsides to this approach, especially in light of 100-microsecond accuracy requirements:
Manual methods don’t scale to the level needed for burgeoning 5G traffic
Testing at remote locations is time-consuming and expensive
Satellite GNSS/GPS atomic clocks are difficult to access and distribute across the datacenter
Cabling of various lengths can insert variable timing errors
Specialized hardware is often needed to get a device timestamp
The need for a more efficient approach to timing and synchronization assurance is clear. But how can datacenter operators evaluate test tools that will verify datacenter timing in live, working networks?
Requirements for a 5G-ready datacenter timing test solution
What capabilities should operators look for when evaluating a datacenter synchronization test solution? Based on Spirent’s ongoing work, we recommend prioritizing selection based on the ability to:
Take accurate measurements on a live network
Support unattended operation
Provide results to a centralized location, such as a network management center
Leverage a built-in atomic clock, such as a Rubidium clock, so measurements can be taken throughout the datacenter without needing to access a GNSS/GPS signal
Identify issues before they impact QoS
Deploy at scale
Of course, there are also technical requirements to balance, such as the need to continuously acquire measurement data, enable simultaneous measurement of multiple synchronization signals, and support Network Timer Protocol (NTP), Precision Time Protocol (PTP) and 1 pulse per second (1 pps) measurements. Importantly, the time/phase measurement must also have microsecond precision and measurements need to be graphed in real-time.
Maintaining accurate time within the datacenter
Learn about our work with a global vendor that needed to achieve sub-nanosecond timing performance for PTP and SyncE. A recent case study featuring Spirent Sentinel explores the challenges encountered and the range of goals that were met in this next-gen, 5G-centric environment.
Guest contributor: Bryan Hovey, Product Manager, Calnex Solutions
Image source: By Clément Bucco-Lechat - Own work, CC BY-SA 3.0, https://commons.wikimedia.org/w/index.php?curid=47366512
