Scheduled Computing Outages: When and Why?
Keeping SLAC's Windows servers current with the latest software updates comes with a small price: once a month the servers must be shut down for a few hours while the latest security packages, service packs, and hotfixes are installed. The scheduled outages, the Wednesday following the third Tuesday each month, coincide with regular releases of Microsoft updates.
"After Microsoft tests their updates and releases them to their customers, we do our own extensive testing," said Jean-Raymond Pierre, Windows Infrastructure Manager for SLAC's Scientific Computing and Computing Services (SCCS). "We make every effort to minimize any problems these updates could have on the lab."
On the second Tuesday of every month, known as "patch Tuesday," Microsoft publicly releases security updates that have undergone extensive internal testing. These updates are installed on a small number of Windows systems at SLAC within 24 hours to look for any adverse effects.
At the same time, desktop updates are released with an installation deadline of Tuesday at 3:00 AM. If users wish to avoid having their systems reboot automatically at that time, they may install the updates at their convenience by double clicking the update icon in the lower right corner of their screen. Any system that has not been patched by the deadline will immediately be updated and rebooted.
The following Wednesday evening, beginning at 6:00 PM, the SCCS group installs updates on the Windows servers, thereby completing the patch cycle.
Most of these updates require systems to be rebooted, which necessitates the monthly outages. And though the updates may only take a few minutes to install, the SCCS group is hard at work during these few hours ensuring that updates have been installed successfully while performing regular hardware and software maintenance and upgrades.
"We make every attempt to complete all our server maintenance in that small amount of time to minimize interruptions to the users and maintain services to the lab the rest of the month," said Pierre.
The outages affect all Windows services including file, email, web, printing, Citrix, instant messaging, and Virtual Private Networking (VPN). The SCCS group suggests logging off, or at least saving and closing all files, before leaving on patch days as any unsaved data will be lost when workstations or servers reboot following update installation.
Ken Kingery, SLAC Today, July 26, 2007