Five Dangers of Poor Network Timekeeping + Easy and Cost Effective Solutions (Part 9 of 10)

In the second part of “providing good network time,” this article will go into more detail on redundant time sources, reliable time synchronization, secure time source, ease-of-use, and cost efficiency.

Redundant Time Sources—A better quality server is one that can receive time from multiple sources, not just one. A redundant time source means the time server can always switch to a different source should the need arise—such as when a company moves or redeploys network assets in a corporate restructuring. Moreover, mounting an antenna to receive GPS signals is not always practical, and signal interruptions are possible. Receiving time from GPS satellites requires mounting a coffee cupsize antenna with an unobstructed view of at least half the sky. Although GPS works with a 180° view, a 360° view greatly enhances reliability and reduces how long
it takes to acquire GPS time.

Reliable Time Synchronization—Of course, once UTC has been acquired, the time server becomes the time source for the network. The way the time server works is as follows: Each of the computers in the network makes requests to the time server for an accurate time stamp. By comparing their local clocks to the time server clock, and accounting for network delays, the local clocks are able to set their clock to match the time server. The key factor that affects the time server’s reliability is the accuracy of its own internal clock. The more accurate the clock, the longer the server can go between UTC resets—and the greater the accuracy between those resets. A rubidium atomic clock (the type used in some of the newer GPS satellites) is the most accurate clock to be found in commercially available network time servers. These can maintain an accuracy of within 1 millionth of a second per day, well within the tolerance of most timedependent software. (By comparison, the Windows default authentication protocol (MIT Kerberos version 5) requires that network domain controllers operate within a time difference of 5000 milliseconds in order for it to authorize logon attempts between the controllers.)

Secure Time Source—Enhanced security is an obvious byproduct of running behind the firewall.

Ease-of-Use—The terms “plug & play” and “set & forget” should both apply to a network time server. Network configuration simply requires plugging the server into the network over a standard Ethernet cable. Setting the time is a one-time operation: you merely plug in the GPS antenna or dialup modem. The server automatically acquires the GPS signals or performs the required dial-out connection. System administrators should no longer be forced to serve as expensive timekeepers and can be freed to focus on other issues.

Cost Efficiency—Time servers are (or at least should be) among the most costeffective purchases that increase the reliability, performance, and security of a network. A single time server costing less than $5,000 can service hundreds of thousands of computers on a network. Compared to other “enterprise” investments, this cost is virtually zero on a per-CPU basis. It is also far below the cost of many time-related operational or
security mishaps.

Get more information on Microsemi’s timing and synchronization solutions now.

The next article in this series will go into redundant time sources.

Read the previous articles in this series:

Introduction: Five Dangers of Poor Network Timekeeping

Overview: Five Dangers and Negative Consequences

Danger: Operational Failure

Danger: Data Loss

Danger: Security Holes

Danger: Legal Liability

Danger: Loss of Credibility

Danger: Providing Good Network Time