UNIX vs NT kernel

[Guest_imported]

Can anyone explain why NT kernel is said to be more complex and bigger than the UNIX kernel?

 

[rycamor]

FreeBSD has an interesting LVM project called Vinum, which allows for hotswapping, and for using non-homogenous disk sets in a logical volume, etc... I think it's also being ported to Linux.

Yes, I hear you about the benefits of some of the commercial Unix distributions. The problem with those is not one of cost, so much as consistency. With different Unixes for different hardware, and different software available on the different Unix distributions, etc... life was just a little more complicated, and harder to coordinate IT across organizations. With open source systems, you never feel restricted. (You won't paint yourself into a corner). There's a certain pleasure in knowing you can just slap together any standard commodity box and run FreeBSD on it, and you can do this from anywhere in the world, without worrying about licenses, complex hardware limitations, availability, etc...

I see the future of application development somewhat like cities and towns in a country: Operating systems and protocols such as HTTP, TCP/IP, etc... are analogous to the roads and streets, while specific software is like the stores where one can buy things. Thus, it is to the advantage of the shopkeepers for all the roads to be free, open, and connected, so shoppers can get from one store to the next. Yes, this promotes competition, so the shopkeepers have to hustle to provide comparatively good prices and quality. Some shortsighted shopkeepers might try to prevent this by barricading certain roads, but they will soon realize that it only costs them more in the end. Certain cities can get away with connecting their toll roads to the main system of free roads, but there is never a situation where a toll road is the only way to get somewhere.

So, to re-explain my analogy, I think in the future free and open source operating systems will be the norm, since it just makes the most sense, while softare to solve specific problems will still most likely be commercially purchased or rented. -------------------------------------------

"Now, this might cause some discomfort..."
(

http://www.wired.com/news/politics/0,1283,51274,00.html)

 

[rycamor]

Oops... forgot to post this great link about Vinum:

http://www.shub-internet.org/brad/FreeBSD/vinum.html

-------------------------------------------

"Now, this might cause some discomfort..."
(

http://www.wired.com/news/politics/0,1283,51274,00.html)

 

[AIXSPadmin]

Thank you for the link about the BSD LVM project. It will be good reading. I have used BSD and Linux before, but have a great deal of experience on AIX and when I use my BSD box at home I truly miss some of the features that I have available. Plus, if I have a cluster of machines I can use PSSP to manage all of my nodes. I know their is clustering software available for Linux and BSD, however, I haven't ever used it and not sure how it compares to PSSP and some of the event management features available like pman and haemqvar.

We did get an IBM p670 (Regatta series) in this week at my company, the regatta series uses LPARs (another feature that is lacking in BSD and Linux; which Sun has and IBM has had for years on the mainframe and now on the RS/6000); but Linux is used on the hardware control management of the p670 I have been told, but have yet to work on it.

I think open source is good and does allow for better connectivity and continuity.

Another topic of discussion would be business continuity and recovery services. We participate in 2 annual BCRS tests a year at an IBM BCRS site. IBM does have the AIX OS and hardware that is needed in a disaster, as well as other hardware, however, if a company decides to go with BSD or Linux or any other open source OS, they would have to have add their OS images to their disaster recovery/backup process so they can recover where they were in an event of a disaster. It cannot be assumed that IBM is going to have every version and release of say RedHat 7.2, and then you get into the patches and how they are going to be backed up and sent with the DR tapes/images. And maybe a piece of software won't work without a certain patch for that version/release of the OS and is IBM going to have every patch (if they support it they will or can get it though) for everything. This aspect is just another piece that needs to be looked at if open source software is going to be used in data centers and businesses. DR is a critical/vital function that must never be overlooked.

Again I like BSD and open source software, and their is really good open source software available, but I do wonder about it being a leading player; sort of like Burger King lets McDonald's do the market research for the best location and then they build in the vicinity (commercial Unix R&Ds LVM, then open source develops as an example). Open source generally, with some exceptions (like IBM backing Linux) does not have the funding and money that commercial companies have for research and development.

 

[rycamor]

Well, all those tools you mention are exactly that: tools, not an intrinsic part of the operating system itself. There's nothing to stop companies from building such tools for Linux or FreeBSD, and even charging for them. In fact, I think a great opportunity is being missed by these companies. Just think how much IBM could make if they would port some of their tools for the BSD and Linux communities. It would be a perfect marriage, in once sense. No one needs those tools unless they have a serious amount of money riding on their data, in which case, I would assume they can afford them. But this would make a perfect migration path for smaller businesses who are expanding. They could start with *BSD or Linux, and never have to drop their use of it as their enterprise data needs expand. In fact, I'm sure this is part of IBM's plan with regard to Linux. -------------------------------------------

"Now, this might cause some discomfort..."
(

http://www.wired.com/news/politics/0,1283,51274,00.html)