For example, we have a database server (behind our firewall) that runs Windows 2000 SP3 and MS SQL Server 7.0 SP4. As of this moment the server has been up... (name disguised to protect the innocent)
"H:\>uptime \\spongebob
\\spongebob has been up for: 198 day(s), 16 hour(s), 22 minute(s), 4 second(s)"
The last reboots were to install Windows 2000 SP3 and a couple other hotfixes, and in the meantime the only service that has been restarted is the SQL Agent, because it stopped sending e-mail through SQL Mail.
My previous database server was Windows NT 4.0 SP6a with SQL Server 6.5 SP5. I wasn't quite able to reach the 180 day mark with that guy, BUT I didn't want to!! My company DID NOT have a firewall at the time, and that server was exposed to the internet! So, in that case, it was very smart of me to install all the service packs and hotfixes very quickly, AND often! Sometimes it's OK to trade stability and up-time for security and robustness.
So when you talk about stability, you have to consider more things than just bugs, patches, and reboots. It's also your ability to recover from problems without restarting the server or its services. I've seen plenty of people confuse database locks (blocked tables) with the need to "reboot the server" ... it's sad really...
|