Windows expert, speaker and author Mark Minasi always offers an interesting spin on the world of Windows. Here...
he shares his thoughts on Microsoft's gold ring; MinWin.
MinWin is a bare-bones version of the traditional Window kernel and operating system that, contrary to some published reports, does not yet exist. Rather, MinWin is a work in progress. The latest versions of Windows for servers and desktops are based on the MinWin work achieved thus far.
"The Microsoft team had a journey of a thousand miles to get from Windows [version] five to MinWin. So far, they have gotten a couple of hundred miles," Minasi said. "MinWin won't actually be real for at least another six years, even though Microsoft is calling their new systems MinWin-based." Minasi likens this misleading label to an automobile company calling a vehicle a hybrid, before the hybrid technology is fully baked.
Why has Microsoft undertaken the MinWin project?
Mark Minasi: There is a general idea that OSes should be modular, like Legos. In the best of all worlds, I could use Windows as a file server and not have to use anything else, but every Windows server is a file and print server as well, because all of that is just built into the OS. Microsoft made the mistake of making everything turn on whether you need it or not, and that means more moving parts, more things that break or have security failures.
In the Unix world, the GUI is an option I don't have to use, so with Linux source code, I can have a lean, mean, fast and cheap OS, and Microsoft wants that. Servers sit on a shelf and tick away, and in the best world, my Web server would not have a GUI, because it isn't needed.
So what Microsoft is doing is this; say you have a car that you only use to go to dinner with your boyfriend or husband or whatever…but it has a tow bar and all this stuff you don't need. Microsoft is trying to take the tow bar and the 12 cylinder engine off the car, to make it smaller, leaner and much easier to secure.
So MinWin will give IT pros a way to use only the components of the OS that they need. Isn't that the purpose of Server Core roles in Windows Server 2008?
Minasi: Server Core today is lean and mean, but you are a little bit stuck with it because you can't dial it up to full server mode. It is like the Smart Car; people like the idea of it, but do you really want to be driven by one?
The Microsoft team wants to be able to ramp up from Server Core to full server or desktop mode, and they have been attacking this. With Server Core [in Windows Server 2008] they ripped down 90% of the GUI, but it is still there. They want to get rid of the GUI entirely, letting people turn off the pieces they don't need, which means turning off vulnerability points.
[Microsoft is] trying to find the interdependencies to get Windows down to the minimal. But the reason these guys are having trouble [getting down to MinWin] is because Windows has been around a long time, and like an old house, it has deep dark corners that no one has shined a flashlight on in years.
When a true MinWin-based operating system is available, how will it help IT pros manage Windows?
Minasi: As an IT pro, I get so frustrated with all of the updates and patches I have to do, which is because everything in the OS is always on. I can imagine a Windows where you put up a basic core, a MinWin, and create a command line, put a GUI tool on there if you want to. Not having to have a GUI means not having [Internet Explorer] - which is the basis of most security patches.
If I am running a Web server or SQL Server serving thousands of people, the fact that I have IE on my server whether I need it or not means that I have to run patches for IE. Comparing Windows Server 2008 [with IE] to Server Core without IE, there are something like 59% fewer patches for Server Core.
Another thing is drivers. Unix, in the late 1980's, developed a microkernel where all the drivers could be put into user mode, which was slower but when a driver crashed, the whole system wouldn't go down…So part of the beauty of MinWin, I imagine, is that you will be able to put drivers into either user or kernel mode. MinWin is really the basis of being able to make these types of fine-grain decisions.
I recently had dinner with [Microsoft Technical Fellow] Mark Russinovich…and he said the whole reason this MinWin idea is exciting to IT guys is because it is guys and their toys: it is a new thing that we will be able to play around with.
Minasi will host informational classes on Windows Server 2008 R2 and Windows 7 this month. His new book, Mastering Windows Server 2008 R2,will be available in January 2010, and he is also working on a book about Windows 7.
Let us know what you think about the story; email Bridget Botelho, News Writer