BigDaddy Posted March 21, 2006 Posted March 21, 2006 I fail to see how can a WINDOWS SERVER have an uptime of 6months?? Since just installing those Windows Updates requires a restart at least once per 2months...
cluberti Posted March 24, 2006 Posted March 24, 2006 Risk vs. reward - either install updates to make sure you have the most recent security patches, or leave the box running for long periods of time and attempt to mitigate security issues via workarounds and security measures.Once hot-patching becomes standard in the OS, these issues should be mitigated due to patches no longer requiring restarts.
nmX.Memnoch Posted March 24, 2006 Posted March 24, 2006 Once hot-patching becomes standard in the OS, these issues should be mitigated due to patches no longer requiring restarts.Is this something new coming with Longhorn Server?
jftuga Posted March 25, 2006 Posted March 25, 2006 This is one reason to use clusters. Have your services running on one server while running updates on the other. Once it is patched, switch the services over and the patch then first server.How to install service packs in a cluster-John
cluberti Posted March 25, 2006 Posted March 25, 2006 I honestly do not know - I doubt it'll be around in the Longhorn timeframe, though. It's more of a pipe dream at this point, but one I truly hope we go for sooner rather than later - I'm not sure how it can be implemented in the current architecture of Windows, but again, one can hope .As to clusters, there are some good and bad situations to use clusters (personal opinion), but if 5 9's are a requirement, cluster is almost the only option that can meet these requirements (as is using 2003 Enterprise or Datacenter editions).
BigDaddy Posted March 26, 2006 Author Posted March 26, 2006 (edited) So how can magazine and surveys even compare LINUX vs Windows UPTIME??Since apperently if you would want a secure WINDOWS SERVER U would have to restart at least once per 2month. Edited March 26, 2006 by BigDaddy
nmX.Memnoch Posted March 26, 2006 Posted March 26, 2006 (edited) Not necessarily. Sometimes you can mitigate the vulnerability by blocking specific ports at the firewall. Other things can be mitigated by local policy (i.e. no web browsing on the servers, no email checking on the servers, etc, etc).The only way that Linux/Unix gets higher uptimes between reboots is that you can usually patch them without having to reboot. The patching systems are completely different. Windows has certain services that can't be restarted with the system being rebooted. The hotfix installation routine can't replace files while they're in use so they're staged and then replaced at the next reboot.Besides that, rebooting for a patch doesn't really count as "downtime". A patch reboot usually takes 10-15 minutes depending on hardware (mainly the SCSI/RAID controllers enumerating devices). If you schedule the patch installations after hours then how is that considered downtime for your business? If you have a 24/7 operation then you should be using clusters as jftuga mentioned.When you put it into the context you're talking about, you don't judge it on how long the server is up between reboots. You judge it on how long the server is up during normal business hours, even if it's rebooted at 3am every morning when no one is using it. Edited March 26, 2006 by nmX.Memnoch
Daemonforce Posted March 26, 2006 Posted March 26, 2006 So how can magazine and surveys even compare LINUX vs Windows UPTIME??Since apperently if you would want a secure WINDOWS SERVER U would have to restart at least once per month.Correct.You have a cluster. It could be two boxes....It could be more. The point is that there is more than one. Virtual machines don't cut it until you need to go back to something.Box 1: Windows Server 2003 Enterprise (DOMAIN CONTROLLER)Box 2: Windows Server 2003 Enterprise (DOMAIN CONTROLLER)Box 1 is the primary domain controller and uses IIS 6.0 as the primary web serverBox 2 is the backup domain controller and uses IIS 6.0 as the backup web server*Update prompt on box 1 and box 2**shutoff IIS on box 1**install updates on box 1**reboot box1*From this point box 2 takes over as a failover serverAll requests made to box 2 are made on box 2When box 1 is finished rebooting, start it up and restart IIS. Check for any problems and fix them before servicing box 2. I remember Service Pack 1 killed the FTP service in IIS and I still can't figure out how to fix that. You might run into a similar problem after an update, so check for any possible problems after a reboot.
BigDaddy Posted March 26, 2006 Author Posted March 26, 2006 So in essence it is still better to be using LINUX! Since my experience with dealing with DEBIAN distro I only had to restart it if there was a kernel change for security reason. Which RARELY happens. But the UI is so much better in Win2k3 server..
nmX.Memnoch Posted March 26, 2006 Posted March 26, 2006 So in essence it is still better to be using LINUX!I don't recall ever saying that. I work for a rather large "organization" and we're nearly 100% Windows. UNIX of some flavor (not Linux) is used only for very specialized applications that were designed by a 3rd party to run only on UNIX. And believe me when I say downtime is something we do not tolerate.
fizban2 Posted March 27, 2006 Posted March 27, 2006 Once hot-patching becomes standard in the OS, these issues should be mitigated due to patches no longer requiring restarts.think restart manager there will be a level of hot patching available in longhorn and vista, but most likely the level we are looking for yet, they are getting there though Carolyn Napier and Tyler Robinson (MSI team) - Installing apps in Windows Vista check out this video on how MSI 4.0 will work with a level of "auto updating" and hot patching utilizing the Restart Manager
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now