Browse Prior Art Database

System and Method for Clock Synchronization in a Dense Server Environment

IP.com Disclosure Number: IPCOM000012438D
Original Publication Date: 2003-May-07
Included in the Prior Art Database: 2003-May-07
Document File: 1 page(s) / 39K

Publishing Venue

IBM

Abstract

In networked systems, typically, each server has their own time of day clock. This requires that all clock must be set individually. This leads to differences of time stamps from the same network. However, is it highly desirable to have all clocks on a network in synchronization, i.e. all clocks have the same time. What is needed is a mechanism that allows all clocks to have and report the same time at any instance of time. This way all transactions to any server on a network that practice the present invention are consistently timed stamped.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 65% of the total text.

Page 1 of 1

System and Method for Clock Synchronization in a Dense Server Environment

       Disclosed is a system and method for synchronizing time clocks in a densely packaged server environment. IBM and other manufacturers of servers offer products known generically as server blades. IBM refers to its offering in this market as Blade Center (TM). Each server blade is a densely packaged server. Each blade has at least a single processor and other legacy PC devices such as the Real Time Clock (RTC).

     The RTC has a time-of-day function and also provides a small amount of battery backed-up storage for retaining information across multiple power-on and power-off cycles. Several such server blades are mechanically packaged into a chassis. The chassis may also contain a management module (MM), hardfile storage, power supplies, a network switch, such as an Ethernet switch, and other devices. Each server blade also has a resident maintenance or service processor (SP) element. Each SP is connected to the SP on the MM via an out-of-band communication or control network.

     On power-up of the chassis, the MM will prevent all blades resident from booting until the MM achieves successful boot. At that point, the MM via its SP will instruct all blade SP's to load the current time passed to it via the out-of-band network, in the time-of-day clock found in the individual RTCs on the blades. The clocks can be loaded with a future time and held off from update operations until all clocks on all blades...