Come on Microsoft!!
When is Microsoft going to introduce a version of Windows to the market that follows along the design of Linux/Unix and Netware's prior upto 6.5 iterations of building an OS that can maintain reliable high availability and uptime based around a single machines for services?Their design based around a kernel that encompasses almost every service into one large file if you will like they have requires far to many patch updates to reboot the server to become effective. I cant believe they have not introduced or at least started to introduce something like NLM's or Linux run level services where a particular netware loadable module or linux service can be restarted for a patch to apply and not effecting everything else the server is doing vs. having to take an entire machine down for a service related update. In this day in age I find it hard to believe they havent been able to come up with a better solution!Yes, this core design philopshy makes 'some' sense, but what you are doing now is requiring even MORE boxes per services to take advantage of this new 'core services' design unless you are running in a virtualized environment.Pleas, someone explain this to me.I long to see the days of my windows servers hitting the 400+ days uptime my linux and netware servers used to achieve with breaking a sweat. Mind you I have some windows 2000 and 2003 servers that are hovering around the 250 day mark but needless to say they have not been patched reglisously. Not every released patch is applicable, yes - but there are some patches those machines really should have. But our environment is large and it is difficult to schedule all these restarts all the time on servers.
March 5th, 2008 9:08pm

This is actually a good question and one I have thought about many times myself. I would be interested to hear some official feedback from Microsoft on why they have never worked towards creating an OS platform in which many of the underlying services and components are individual allowing restarting of single services rather then the entire system.
Free Windows Admin Tool Kit Click here and download it now
March 7th, 2008 5:14pm

scott_k2003 wrote: When is Microsoft going to introduce a version of Windows to the market that follows along the design of Linux/Unix and Netware's prior upto 6.5 iterations of building an OS that can maintain reliable high availability and uptime based around a single machines for services? I really find it hard to see the purpose of long periods of uptime. I have always asked myself why linux, unix and Novell persons havea fetish with uptime and compare it betweentheir and other Operating Systems. The questions that are really important are: Is downtime the real problem? (or is it budget, knowledge, culture, etc) Is scheduleddowntime really a problem when service windows exist? Are clustering and virtualization adequate solutions to the availability problem? Can this problem best be solved using a technology? Can communication, planning and other soft skills be used to work around or solve the problem? scott_k2003 wrote: I cant believe they have not introduced or at least started to introduce something like NLM's or Linux run level services where a particular netware loadable module or linux service can be restarted for a patch to apply and not effecting everything else the server is doing vs. having to take an entire machine down for a service related update. In this day in age I find it hard to believe they havent been able to come up with a better solution! Actually Microsoft has recently introduced features in this context: Restartable Active DirectoryWindows Server 2008 Domain Controllers are the first domain controllers with an Active Directory that can be restarted. More info. Restart Manager (codename Freeze Dry)Microsoft has worked on 'freezing' parts of the Windows Operating System to limit the number of reboots to the absolute minimum.The technology (present in Windows Vista SP1 and Windows Server 2008) can patch parts of the Operating System as long as no critical system services are involved. Drivers being increasingly more user mode instead of kernel mode will also help make the experience of rebooting every second Tuesday o each month a thing of the past. More info. When I look back I see Windows NT 4.0 servers that needed to be rebooted when you changed the IP information, werefilled with kernel mode driversand servers that needed to be reinstalled when you wanted itto change tobecome a PDC, BDC or Member server.Some changes require the help of the ecosystem though and changes aren't always easy or quick. scott_k2003 wrote: But our environment is large and it is difficult to schedule all these restarts all the time on servers. In large environments virtualization and clustering can make sense in terms of availability. In a cluster environment a cluster node can be taken offline for patching while the other node steps in to keep the cluster resources running.I believe a business that requires a high number of 9s knows this requires the budget to grow accordingly. The same goes for Confidentiality and Integrity. (principle of the CIA triad)
March 7th, 2008 11:42pm

Hello Scott, You should also read through the information on the Server Core installation option in Windows Server 2008. This installation option will greatly reduce the number of services/features in the Windows install to just whats needed to run critical roles. This will reduce the number of updates you will need to apply to these systems. This was the idea behind the option. Every iteration of Windows Server has made great strides to become more modular and administrator friendly. Hope this helps, - Justin
Free Windows Admin Tool Kit Click here and download it now
March 11th, 2008 10:09am

This topic is archived. No further replies will be accepted.

Other recent topics Other recent topics