Home Server Projects
I've always been interested in installing servers for home use. The ability to customize Linux based servers to the degree that surpasses what is economically available from any other source makes Linux the obvious choice. Linux also offers the home user more latitude in that it will support practically any other desktop operating system with basic services.
Sunday 27 May 2007, 6:01 AM
A challenge to all Linux Users in the World
A recent posting by George Ou about Windows Home Server brought out a lot of responses from the Linux community. I think that Microsoft deserves to get some competition.
Here's the challenge:
There are a lot of Linux experts on this site. I challenge them to come up with a list of applications that work the best for each of the given functions found in Windows Home Server. Show me a list of something that you guys think will beat or meet Windows Home Server, that will work for a lot of people and that you think can be easily reproduced. I will build it and give you an honest appraisal of my experiences while doing it. I will document it well enough that others less skilled will be able to do it. I'd like to be able to release it as a DIY compilation distro if possible.
Now the kicker, on the EXACT same hardware I will install the WHS Beta or release RC1, assuming I can get it, and test it in comparison on the basis of a USER, not a technician or a systems engineer.
There are a few restrictions I place on this challenge:
1) Do not expect me to compile Gentoo or Debian. I want to get this all done within a few weeks, OK?
2) The install packages must be applications that can be downloaded from the distro's website or mirror. I will not start with somebody's forked code.
3) The applications need to be compatible with the Linux distro and the desktop. I am partial to Gnome but I'm willing to use KDE.
4) If you expect me to script something or run a script, show me an example or give me the script. (This runs counter to item 2 above but I realize Linux runs on scripts). It would be nice if the script has enough commentary to let me know what was going on in it. I reserve the right to dump something I'm suspicious of.
5) If the application mix doesn't work, be prepared to get bad Linux PR from the blog. I will not attack or flame anybody personally as a part of this challenge but the distributions and/or applications are fair game. If they're bad they need to be flamed.
I will take a complete list of Linux applications from an individual or a committee of no more than 3 people and a suggested Linux distro to put it on. I will build it on a 2.6 Ghz Celeron system with 512 MB Ram on a stock DELL with a CDRW or a DVD +/- RW drive. Since this system will NOT be playing the DVD or video, we'll live with the on-board SVGA chip and the optical drive is there for software install only. (Its a server remember?) I have my choice of hard drives.
Then I will report back blog style what it took me to do it. I'll keep track of the hours and what and where I had to find my manuals/documentation to make it work. The idea is to develop a "distro" that can duplicate the published capabilities of the WHS minus the bare metal recovery. I can use Ghost as well as anybody.
My credentials:
I'm not a Linux expert but I have messed with it a little. I have 2 SUSE 10.1 Linux boxes and 2 XP Pro boxes running at home. One of the SUSE boxes will be used for the challenge with a blank drive. (I'll save my current drive for later.) I have Ubuntu 6.1 and 6 or 7 Windows CPU's in my cubicle at work. I have run various Linux distros on desktops since Red Hat 6.4. I've tried Mandrake, Knoppix, Storm, Yellow Dog, Red Flag and others I've forgotten about.
I design Windows XP Embedded systems for a living. I've been doing that for over 6 years. I started with Windows NT 4.0 Embedded for 2 years before that. My system designs run in mission critical environments in the oilfield every day all around the world on over 300 systems. My first programming language was Fortran on punch cards. My first personal computer had a S100 bus. I've run C/PM, VMX, LDOS, MSDOS, Windows 3.0 through Windows XP. I've programmed in Z80 and x86 assembler, Forth, about 15 Basic variations, Algol, C, VB Script and lately I'm learning some of the Visual Studio 2005 languages.
Do I like Linux? I like the idea of Linux. I like the idea that there is some competition for Microsoft. I haven't found a distro I like enough to hunker down and spend a lot of quality time with it. I do plan to have a full suite of auxiliary servers running on the SUSE box that's not involved with the test for DNS and DHCP. I also will be running network sniffers to track what traffic is traveling on the subnet to and from the client and the server. Microsoft is notorious for using a lot of “silent” traffic to and from un-documented ports.
Remember we're looking at a Home Server as an appliance not a high performance server. Installing it on the exact same hardware mostly removes the hardware performance issues from the comparison. Obviously how well device drivers were written for the two operating systems is not removed from the test but at the level I'll be testing, device drivers will just be lumped in with the OS issues.
So the test criteria, in no particular priority, are:
1) Does the software adequately support the feature set?
2) Does it work well and reliably enough that home users can operate it successfully?
3) Can repetitive maintenance functions be automated?
4) After running it for awhile, are there obvious gaping holes in the features offered that need to be filled?
5) How easy is it to install? Can it be customized as it is installed?
6) Can you add features to it after installation? How easy is it?
7) The question of updates, ease and availability.
8) Discern the need for additional services that aren't a part of the system.
PreviousComments on this post
What can Home Server do for you? I'll tell you what Linux Terminal Server Project (LTSP) does for me. You assume everyone reading this knows what Microsoft Home Server enables a person to do. I have not a clue. What you don't seem to understand is that you assume that all Microsoft allows you to do is all you would want to do. Humorous to a long time (non-geek) user of GNU/Linux (Linux is the kernel. GNU/Linux is the OS). I also find it humorous that anyone would use Microsoft products in a server capacity. Server capacity was an add on afterthought to Microsoft OSes. At work I must endure Microsoft products. At work this year we got a new (enterprise) server/thin client system that cost in excess of $40,000 US for server, twenty thin clients and the software. Using it, I am too often frustrated in userland where I find one functionality after another eliminated in an attempt to secure the system. UNACCEPTABLE! The system is junk. The system is pathetic in comparison to the LTSP system the Free Software Association I belong to installed in a local community center. It has fifteen thin clients with local sound and removable media access and far more application programs than are available at work. The total cost for this LTSP network (GNU/Linux) was $3,500 US plus volunteer labor. Adding more thin clients would be cost free. We used a new computer as server and junk computers as thin clients. It works much better than the system at work. I could probably set up a LTSP thin client network on my own by following online directions and googling solutions to any problems that cropped up. However, when I set up my home network I paid a GNU/Linux wizard for an hour to set the software part of my system up. My system: Silent thin clients with local sound and removable media for each member of the family; powerful noisy server isolated in the cool basement; printer in my office; projection video (DVD, etc) in livingroom (we don't have TV but if I was interested in TV I would have a GNU/Linux wizard set up Myth TV so I could automatically record and time shift TV programs and eliminate commercials). Data backup is automatic and off site. As with any GNU/Linux box, I can securely SSH into my user account (or the root account since I am the system administrator) from anywhere in the world. (Last I saw the software to that with a Windows box cost a lot of money.) My GNU/Linux (Ubuntu, but I've used Debian and Fedora in the past) system is of course virtually immune from viruses and spyware. No user can affect the server or any other user in any way. (Unix and Unix like systems are inherently secure because root and user(s) are firmly separate. Things done in userland do not affect root.) I have a secure, fast, stable system that meets all my needs. What can Home Edition do for you? Could you hire a hacker to add functionality that Home Edition doesn't come with out of the box? With LTSP I can (legally) add whatever functionality might need. I just have to have the skills or money to pay for the skills to have that functionality added.
This comment has been deleted at the users request
You do raise a valid point. I should have detailed the features of MS Home Server.
The features of most interest are automatic backup and bare metal restore of the client. Using thin clients more or less automatically eliminates the need for backup since its all on the server. Likewise bare metal restore is also a moot point. Automatic backup is actually trivial in both MS and Linux installs. AT or CRON takes care of that. Bare metal restore is Symantec's Ghost or numerous other applications.
Another feature is journaling of storage transactions which various flavors of Linux support natively.
Single copy storage with linking for subsequent copy events is another touted MS feature.
Using RAID features for the actual storage medium is another feature of relative minor note. Probably easier and more reliably done in hardware no matter the OS.
What I am mostly interested in is: What does it take to build a media storage and playback server using Linux and the applications that run on it that can meet or beat the general consensus description of a media centre/server?
I'd like to start friendly discussions with people who have done it already and see what they think is the best methods and applications to do it. Thin client is one means I agree. My family uses Windows and Linux. I'd like to support both from a Linux Server.
If you have suggestions I would greatly appreciate knowing your thoughts on the subject.
Eventually the goal really is a composite distro that when installed will do it all for a home environment. I'd be very happy to publish it back on the web.
I would recomend you listen to linux reality, a podcast for the new linux user.
In episode 62 he discussed music servers, the easiest to set up was the firefly server http://www.fireflymediaserver.org/ which is a dapp server (itunes audio)
http://www.linuxreality.com/podcast/episode-62-home-servers-part-8-music-servers/
Episode 63 is also useful where he discusses a variety of backup tools.
http://www.linuxreality.com/podcast/episode-63-home-servers-part-9-backup-servers/
I'm a bit puzzled by your challenge proposition by some points even thought the idea of making this comparison is a good thing.
You plan to use a celeron cpu and use gnome as window manager... good idea if you plan to waste a good part of the power of cpu that is knowm for being quite limited already. There are lighter gui's around in the linux world & they should all be sufficient for what you plan to do, eventhough i'm surprised an embedded systems specialist needs a gui on a server.
Considering your experience in programming i'm also surprised by your dismissal of gentoo & debian solutions, which provide the largest repositories of applications that are available and are probably amongst the top3 most versatile too (with Slackware i think).
Your comment about not being able to set up that media server in a few weeks using one of those has no grounds... If you really plan to make a composite distro they're probably the best choice, unless you want to try out the Linux From Scratch way...
"I will take a complete list of Linux applications from an individual or a committee of no more than 3 people and a suggested Linux distro to put it on."
First you want advice from everyone & then you want to use a list from only 1 to 3 persons???
"Obviously how well device drivers were written for the two operating systems is not removed from the test but at the level I'll be testing, device drivers will just be lumped in with the OS issues."
That won't give any idea about the quality of the 2 systems... unless all the server hardware is supposed to be supported by both OS's, which i guess must be a prerequisite for a valid comparison.
"What I am mostly interested in is: What does it take to build a media storage and playback server using Linux and the applications that run on it that can meet or beat the general consensus description of a media centre/server?"
Given all the other features you talked about are supported from the box with most major linux distro that leaves us with this one, maybe you could elaborate a bit more about the features you're expecting.
You may also want to have a look at projects like GeexBox, Myrinix... well there are so many around...
Looking forward to see what will come out of this...
This comment has been deleted at the users request
The reason for eliminating Gentoo and Debian is that I want a system that can be built QUICKLY in 2 to 3 weeks. I have no doubt that Gentoo or Debian distros can eventually be used. And yes I forgot about Slackware. My experience with C is so far in my past as to be ancient history. I do not want to have to rebuild my programming skills then tackle the applications that I am sure are more likely more mature than I can accomplish in a few weeks, that being defined as 2 to 3.
What I meant by committee selections would be a group of 2 to 3 in EACH group. I've spent a lot of time in meetings with programmers, if the group gets bigger than 2 or 3, nobody can agree enough to make a single suggestion. Usually an individual can cite at least his favorite application and has reasons why its his favorite.
The lightest GUI involved would be a Web browser on another computer, the only use for the Gnome or KDE would be for the initial install and setup. After that I hope to yank the monitor, keyboard desktop and mouse off.
Oh brave Anonymous, your nit-picking observations like others are important, however, you have fallen into the same rabbit hole I have found already. Lots of suggested applications, no real How-To's, no system critiques, no examinations or comparisons between one Linux application and another. A lot of dither hither and yon but no definitive out-right declarative statements saying do this because its the best for this reason __________ (fill in the blank).
I have done a certain amount of research on the subject of Linux Media servers and the 2 systems you name I've yet to come across. The problem is that there are hundreds of them. Yes there are so many around, exactly my point. Yet none of them seem to have gathered a large enough following to make much impact outside their group of enthusiasts. Myth TV gets some interest as a TV programming component. There are a number of mp3 players/recorders etc. I'm trying to assemble enough reasonably functional applications that are compatible to make an appliance as opposed to a geek-toy.
That is the area where I've seen the Linux community fail to understand the general user population. They like to make geek-toys and they don't understand that they COULD build a giant-killer. IPod is an excellent example of a potential Giant-killer. Google Earth is another. Both sharply focused applications that can be expanded in purpose and market target.
I would like to see your definitive suggestions. I'm assuming you consider yourself an expert?
As to the reasoning behind the choice of a Celeron CPU. That's easy, that's what I have available for the test. Not to mention the fact that as a server it doesn't really need much video output.
The second test running the Windows product is solely for comparison testing, if you noticed the test criteria the focus is on USER features and ease of use not performance. I'm sure Windows will be trounced there.
Dear xMSjunkie ;-)
there's a big probability that all the apps you need are already there (considering you still haven't detailed exactly what media capabilities you're looking for), as you suggested there may be a couple of scripts to write if you're not completely satisfied with the integration you get but that shouldn't be too difficult for a man of your experience. So you shouldn't have any need of updating you C skills...
Once again there's no reason to think debian or gentoo would take more time to setup & if you really want the easy way, pick up a derivative that's already configured and provides an easier setup & some more bloat, strip out want you don't need and add what you do (f.e. Kanotix. Sidux, Sabayon Mini edition)
"Oh brave Anonymous, your nit-picking observations like others are important, however, you have fallen into the same rabbit hole I have found already. Lots of suggested applications, no real How-To's, no system critiques, no examinations or comparisons between one Linux application and another. A lot of dither hither and yon but no definitive out-right declarative statements saying do this because its the best for this reason __________ (fill in the blank)."
As you maybe know it's not possible to submit a post here without being registered. So, i don't know why i'm listed as anonymous, but just to clarify things i'm the brave ShakaZ.
Well i'm can only say i'm astounded by your statement, all those missing things i happen to find in abundance with my friend google. Of course you may have to do a bit of research to find the quality of information you want, but that's the internet & it's the same in windows world...
A good place to look for decent howto's : http://www.howtoforge.com/
Other places to get info or advice : the distribution's & software project's websites, forums, irc channels, wikis,....
There are of course many articles where one states why one should use this or that application or settings. However in an opensource world there is choice and not everyone will agree about the best apps for a particular task, there are though general trends that appear and many distributions do actually come with a lot of software in common.
Only thing you have to do is try out different software and see which you prefer.
"I have done a certain amount of research on the subject of Linux Media servers and the 2 systems you name I've yet to come across. The problem is that there are hundreds of them. Yes there are so many around, exactly my point. Yet none of them seem to have gathered a large enough following to make much impact outside their group of enthusiasts. Myth TV gets some interest as a TV programming component. There are a number of mp3 players/recorders etc. I'm trying to assemble enough reasonably functional applications that are compatible to make an appliance as opposed to a geek-toy."
Well now you know about them why not give them a spin and see if they could be a base for your project?
If you want better advice you should specify what media capabilities you want... streaming audio/video, remote control, terminal server, file server...???
"That is the area where I've seen the Linux community fail to understand the general user population. They like to make geek-toys and they don't understand that they COULD build a giant-killer. IPod is an excellent example of a potential Giant-killer. Google Earth is another. Both sharply focused applications that can be expanded in purpose and market target."
In fact Linux is already a giant killer. It leads the server market & is gaining desktop users everyday, several governments and big companies have already made the switch which gives an idea of the quality linux has achieved on the desktop front these last years.
"I would like to see your definitive suggestions. I'm assuming you consider yourself an expert?"
I have none given the information available & as i'm by no means an expert i wouldn't consider my suggestions definite. I just happen to be a free software user for almost 10 years.
Sorry about all my trouble posting.
There is a Ghost program (gghost or something) for GNU/Linux but no one uses it because there is no point to it the way there is with Windows. Command line works better. Besides, installing most GNU/Linux distros is can be done in less than an hour. Do you really need a Ghost type program at home? A new GNU/Linux install is fun, "Maybe I'll try this ap, and this one, and this one...".
GNU/Linux does raid just fine. But raid is not a good back-up choice. There are other reasons for using raid.
GNU/Linux is much better at serving mixed networks than Windows servers are. If you are going to have a mixed network definitely use a GNU/Linux server. SAMBA is said to server Windows boxes better than Windows own server options. Depending upon what the Windows boxes are to be used for, it might be possible for you to run Windows in a virtual machine on the server. The thin client network could then allow all users to seamlessly use the GNU/Linux and the Windows. Theoretically. I don't know of anyone who has actually done this. Things run slower in a virtual machine in any case. If it's not the Windows environment, but simply a couple of Windows aps that are desired, they may run in Wine or another emulator, and quite possibly run better in the emulator than natively in Windows (a friend found this to be true for a particular CAD program he was using). And don't try to play graphics intensive games or do other processor intensive work on a home network thin client. There is plenty of information on the net and in book form for setting up and administering SAMBA. I encourage your family to kick the Windows habit and join the Free world of Free Software. Dual boot if you must for gaming or the few other things better done on a Windows box, but make the switch to GNU/Linux. Freedom matters more than temporary convenience.
Check the "Linux Journal" archive. The magazine commissioned a media center computer a year or two ago, maybe three, and did a review/report about the outcome. For whatever year it was, it was the Ultimate Linux Box LJ does each year.
Your reluctance about Gentoo I can understand, a bit. It takes forever to compile everything, though you don't have to watch; you can sleep or go to work. And Gentoo has a steep learning curve (too steep for me), though you will learn the ins and outs of GNU/Linux in the process. Reluctance about Debian is silly. It is as easy to install as any distro, or nearly so. I didn't have any trouble the several times I installed it. Debian's problem is Stable has ancient versions of everything. Testing is current, but may need massaging by an expert sys admin. Ubuntu is a better choice for people like me. Now that I understand your desires better, I agree with another poster that Debian, Slackware, or Gentoo are probably your best bets. It wouldn't surprise me if Gentoo was your best bet for getting what you want. However, I just did a search of Ubuntu Synaptic package manager and Myth TV is listed. Wouldn't it be fun if it just worked upon download? :-) Might. Mind the TV tuner hardware you buy and make sure it will work with Myth and GNU/Linux.
Even if some parts of a GNU/Linux media server (that seems to be what you want) may not be as easy to install and setup, or to use, and may not even do all the things a Windows system will do, the GNU/Linux system will do things that the Windows system will deliberately prevent you from doing. When you make your judgments comparing the systems take that into account.
For a fair comparison you must make sure all the hardware has fully functional GNU/Linux drivers available. Since 2001 when I switched to GNU/Linux I have done this, and not had problems. Drivers are less and less a problem with each passing year, but graphic drivers are a problem, especially for distros and individuals who are not willing to use non-Free (as in Freedom) software, including drivers. The driver issue is a long discussion and I'll simply say the problem is not GNU/Linux, it is hardware manufacturers.
Unless things have changed recently, Myth TV is a pain to set up. But it does work well once it is set up properly. Best to hire an expert who has had success with Myth, and consider it a dollar cost of using GNU/Linux. And buy your hardware according to what will work with Myth TV. ("But then the hardware will cost more." Maybe, and there will be the Sys Admin cost if you hire someone. The Free Software I use for all the various things I do with a computer cost me $0 where-as the equivalent software for Windows would cost me in excess of $5,000 US. I don't mind being picky about the hardware I buy, and don't mind hiring an expert when needed.)
Given that you are a GNU/Linux newbie despite some experience using GNU/Linux, and given the fact that what you want your system to do is ultra geek, power user, your most reasonable course of action would be to order a system built to your specifications of hardware and software capabilities from a company specializing in building GNU/Linux systems that work out of the box. Otherwise, make a serious commitment to get very familiar with GNU/Linux. Get friendly with a GNU/Linux wizard.
This comment has been deleted at the users request
Now you are unfairly criticizing the world of Free Software.
There are tons of choices of software because there are lots of individual choices about what one wants to do and how they want to do it; your's being one of those choices. Some of the choices are junk and sometimes there are multiple choices of excellent software, most frustratingly sometimes I want to choose this capability from this software and that from this other one. Guess what? I wait a few months and quite often each of the software choices has the capability I wanted. I have learned to choose the software I am most comfortable with even if it doesn't have a particular function another has and I wish it had.
I am definitely NOT an expert. I get help administering my own system. But even if I were an expert I could not give you the information you ask for. You still have not stated what you want to do with enough specificity for anyone to do more than we have done, point you in the general direction and make a few general suggestions about appropriate software. I suspect you will not be able to be as specific as required until you have seriously sorted through all the various options and seen the capabilities that exist.
I'll warn you that your killer ap or killer system notions are naive. I speak from experience. I thought LTSP was the killer ap: buy the best computer you can afford, use the old ones as thin clients and every member of the family gets to use the best computer in the house. No one else has been interested. Naive does not mean you are wrong. You may be correct. The system you want may have mass appeal. Design your perfect system, build it, modify it to perfection, then produce a few systems and see if you can sell them. If you can, you may have the next ipod, and your brand name will be the system everyone else is copying so you will have a marketing advantage.
Hehe double post again freedom ^^
I agree with most of what you said, except for the part about running windows in a virtual server as that would drain way too much resources & i'm not even sure if it works if the server runs without gui...
Other media center projects you may want to look at apart from the 2 in my first post :
Elisa Media Center : http://elisa.fluendo.com/features/
Freevo : http://freevo.sourceforge.net/about.html
Doesn't look like geek-toys to me o_0
Lets get focused here. The intent is to assemble a group of applications for media record and playback from a server that will NOT be used for the actual playback of the streaming media. It will be a Linux file server that can capture TV signals either Internet, cable or satellite sourced, it can download music files, podcasts, etc. and store all of it in formats compatible with either Linux or Windows clients. The render engine for the media will be on the clients, not on the server.
As to paying somebody to build the system, the whole point is to engineer/build the system and then document what has to be done to make it work as a whole system. If I wanted to have somebody just deliver the blasted thing, I could buy it from Microsoft via freaking DELL.
Why don't you take a look at the solutions we provided, the media center distros i mentioned offers those capabilities already...
So either you pick one of those and set it up with the right hardware & eventually add a couple of software for functionalities that would be missing, either you build your own system using another distro as base and inspired by the parts you liked in those projects...
Once you have your system setup you can make a livecd of it and share it with whoever you like...
For disk/partition cloning you can use partition image (partimage) for example, only thing is the partition you're backing up that way must be unmounted so you may have to do some hacking if you want that to be an automatic process, but i'm pretty sure this problem has already been resolved... There is also linux software doing incremental backups, but you'll have to look that up as i'm not using any of those.
You're an idiot. You're a moron. Typical Windows/Microsoft fanboy. The ultimate solution: stop being an idiot, or moron. Maybe both.
Here's a challenge to Windows' (of any kind) administrators and Windows:
Have the Windows server send an email to user1@host1.com when your server detects an email which is being sent to user2@host2.com from user3@host3.com.
This challenge is a one liner in any linux/unix operating system and could be setup in minutes, and no extra software need be purchased or installed.
I guess the questions boil down to this:
What have you used that you liked for media functions on your systems?
What applications do you suggest I stay away from? Or that you've had problems with?
I will look at the suggested distros and the applications.
In response to ugarit:
In Exchange server its a couple of clicks in an management window.
To do it in VB script, requires 3 lines. The first line creates the object and passes the credentials to the remote server. The second creates the event handler. The third line sends the email.
On a single workstation in either a work group, Samba domain, Active Directory (that's LDAP managed domain to you Linux guys), or NT Lan Manager domain, you don't need to use email, Messaging is built into the OS. And yes I can send a message to my Blackberry or PDA or phone if I want without writing any scripts, or buying anymore software.
So what? All modern operating systems include similar capabilities. I could do that in C/PM, 15 years ago.
The purpose of my challenge is to ask people who are familiar with Linux much more than I am to offer me information on what has worked for them in the realm of multi-media services and applications. I'm not interested in getting into a pissing match between MS and Linux adherents.
In response to Xwindowsjunkie:
You did not meet my constraint (which I wasn't clear about) and that it be done without installing extra applications. MS Exchange is certainly an extra one and quite expensive. Clicks are time consuming.
You're right the capability existed in most OS's for decades and that was my point.
Don't worry, I don't participate in pissing matches.
I agree that Exchange Server is expensive but technically its still one BIG MONSTER mail server. I used to have to manage an Exchange Server and I hope I NEVER have to do it again. Hell for me is having to deal with 5000 email clients all PO'ed that they can't get their email on Monday morning.
I have downloaded all of the suggested distros and applications I have gotten suggestions on. We'll see what happens once I burn the CDs and try them out.
PM 5/28/2007
It looks like though I will need to do a lot more looking around. Some of the suggested support Linux quite well. I'm going to have to work a little harder to find a means to support Windows clients on the Linux box.
I hope you are able to come up with the system you want. I think your present course is one leading toward success.
Here are links to the Linux Journal covering the Ultimate Linux Box media center. The info is less than a year old so still relevant.
http://www.linuxjournal.com/article/9035
http://www.linuxjournal.com/issue/148
Let us know whether or not you meet success.
X,
I understand what you are proposing, and before I question its validity or usefulness let me suggest Ubuntu Server as an out-of-the-box, under 500MB download solution to what you seek.
Given the user-orientation of desktop Ubuntu and its derivatives, the server version which installs in about 20 minutes, or less depending on hardware and bandwidth, and given how drop-dead simple adding all the functionality you seek if it should be missing in the default configuration: e.g. "sudo apt-get install dhcpd" if you decided that you should have checked that option in addition to LAMP among the few questions asked during install, I'd assume Ubuntu Server will pass your test.
Now, to circumvent any objection relating to non-geek, typical home user try Synaptic - yes it's included. Ask to see System software or search for DHCPD by name and click play.
Tell me that you've XPerienced anything smoother, and more brain-cramp free in the Wonderful World of Windows than this initial install and application installation. How about time to complete? How about upgrading applications or entire sub-systems? (allow me the CLI: "sudo apt-get upgrade Apache2") or upgrading the whole damned distro ("sudo apt-get dist-upgrade").
HOWEVER, my quarrel with what you are proposing here, attempting to accomplish, and motivation explicit and implicit is who gives a shit?
When I am asked for a recommendation do I cite cost? Do I cite TCO? Do I cite security? Do I cite numerical figures for application superiority/inferiority? Performance? Eye candy? freedom? Freedom? Et freaking cetera?
Maybe, though often not.
What I am most likely to cite, depending upon to whom I am speaking is my pure, undiluted, sustained enjoyment and satisfaction.
Since 1998. Hardly first on-board, but that's pushing ten years.
I've used GNU/Linux, I've used Windows largely at work and rarely at home, Solaris, Macintosh OS X, but I can always make my highly efficient and productive work happen as I choose and prefer right here.
Casual user? Modest equipment? No, developer, database and server administrator, programmer since 1980 in assembler, Forth, and most of the popular low and high-level languages that followed. Machine in back bedroom - SMP Dell 410 Workstation, busmastering SCSI subsystem, high-end graphics, etc. Study/Computer room - SMP Dell 470 Workstation 2x3.0GHz Xeons, 1GB RAM, 2x10,000RPM SCSI drives under PERC320 busmastering controller, nVidia Quadro workstation graphics, etc. Laptop - MacBook Pro Core 2 Duo 2.16GHz, 2GB RAM, 120GB ATA drive, Parallels Desktop virtualization (beta testing v.3.0 at the moment).
Do I wring out my hardware, my OS, my experience? Uhh...yes!
So, and I'm truly sorry to have run on this far since what I primarily wanted to assert is that you're wasting your time and any results you achieve will be largely meaningless.
I mean...what's cooler? Aero or Beryl/Compiz? Who gives a shit. Which has more apps and in what arbitrary category? Windows? Ubuntu? Who gives a shit. Which OS/desktop/environment supports real work, not to mention home server utilization and administration...by people of the lowest experience/intelligence/determination. Man, who gives a shit.
How much would I miss the Vista seen out the Windows and its Genuine Advantage and thrilling XPerience, and the warm fuzzy protectiveness of User Account Control - no I don't mind being interrupted by pop-ups, challenges, freaking Wizards, multi-paragraph dialog boxes, stream-of-unconsciousness insults to my intelligence (don't you dare say that OS X is even more childish and brain-dead until you try it and see what it's like to work uninterrupted, and double don't you dare say that a CS degree is required to wring every drop of productivity out of GNU/Linux and finish feeling accommodated and supported as opposed to cattle-herded and programmed until you've tried it too)?
Heh...heh - ohhh, terribly.
Would I choose Windows Home Server Grande Deluxe Pro SP2 because it had an application or two that its advocates deem "best of breed?" Ever been to a dog show and seen some of those breeds?
No, I wouldn't.
Hey, good luck X with your little demonstration. I hope you either prove your point, such as it is, or learn something in the process.
Frank
p.s. Wanna soil your saxophone?
Take Ubuntu Studio out for a spin, then sit down and calculate what it would take - and go ahead, plot availability and functionality - of duplicating that warehouse of wank-o-rific wonder in Windows. Be sure to factor in the obstruction, interruption, DRM, treadmilling, surrender, risk, Fear Uncertainty and Doubt about what you've gotten yourself into.
How many trips to how many stores and how many hours of configuration in getting even most of it to work together and how much study to become a "typical user" in this highly-specialized area, and how much flexibility, fun and satisfaction when you discover your first Minum Opus can't be shared with your girl friend without circumventing the it's-for-your-own-good DRM and breaking The Law?
http://g-ding.tv/?q=MythDora
fpirrone :
Actually I do not disagree with your commentary. Even if I manage to make it all work even with Ubuntu 7,04 Server Edition (which I downloaded and burned to CD today) there are still user interface issues. And if I succeed in documenting it all and nobody uses it, so what. I will have learned a lot about Linux in the process. I'll have learned even more what it takes to have a mixed bag of Windows and Linux boxes working together. I'll have added a new skill set to my overloaded experience with Windows.
One of the things I do for a living is to design UIs that even high school dropouts with no computer experience can use in the oilfield WITHOUT trashing the OS. To a Linux user it might seem weird that a UI has to be designed to keep the user from damaging the machine but that really is an issue with Windows unfortunately. More often than not an oilfield worker gets tired of waiting for Windows to do its shutdown dance and just pulls the plug to shutoff the computer. Plus no matter how much work you do within an application there are a number of little utilities necessary to put the finishing touches in.
That's sort of the approach I'm using here. What applications provide the necessary functionality, with a user interface that allows the user to ease right in and do what he wants with the system. Then still have the finishing touches to cover the items that probably could be done if the user was a little more adroit with the OS but isn't. In other words, make the software and hardware operate as easily as a washing machine or a VCR. Ignoring the clock that nobody seems able to set properly!
Considering the press I've read especially from a few of the bloggers I trust and admire, Ubuntu 7.04 Server sounds like the real deal. I'm going to try and get to it Thursday or Friday.
As to the issue of DRM. That is a fear I have already and I've refused any Windows Media Players past version 9 that come down the pipe because it seems as if its likely one day that video I create on my own equipment will not be allowed to play because I haven't bought a license from somebody. I see that happening already with DVDs. If I make a home movie I can't play it on somebody else's equipment if I happen to attempt to play it on a computer from some other DVD region. And that's just flat out wrong. That's restraint of trade and censorship.
>"I challenge them to come up with a list of applications that work the best for each of the given functions found in Windows Home Server."
Hi there,
Can you point me to a feature list?
>1) "Do not expect me to compile Gentoo or Debian. I want to get this all done within a few weeks, OK?"
I am not certain that you understand "Gentoo", or Debian for that matter.
All that said, I could probably crap out a custom live CD or base image in 72 hours, given some starting criteria.
Thank you for your time,
Frank Russo
I've been running a "home server" for close to 10 years, and I think one of the first things needed is to simply define the term. Starting with "Microsoft Home Server" may not be the best way, because it starts with THEIR definition, which may or may not be a good one.
To begin, my home server is my most recent cast-off from the deskside. As such it's by definition "low power". For both performance and security reasons it doesn't run a GUI, so GNOME or KDE simply doesn't matter, because the answer is "none of the above, nor xfce or anything else." If I weren't comfortable with ssh and a command line, I'd likely use webmin, or something like that.
Functions on my server:
* DNS, authoritative for local, caching for the world.
* DHCP, "known" machines get fixed MAC-IP mapping, visitors from a pool of dynamic addresses.
* IMAP, for incoming email. When I get more time, I want to use "archivemail" from a cron job to automatically move old email to separate folders, eventually to compress, and someday delete.
* Postfix, for general email function, and outgoing email.
* leafnode, for news. (Not currently fully functional with new ISP) I also bridge mailing lists into local newsgroups.
* NFS, for the same $HOME no matter where in the house I am, or what machine I'm on. In addition, the NFS data is on a RAID1 mirror, for some measure of hardware backup. When I get more time, I want to run "cp -al" from a cron job to do nightly link-only backups.
* Someday I'l like to run LDAP/Kerberos for single-system-image login, etc. But that takes even more time than what I haven't done already. Woodworking and yardwork come first.
I also run mythtv, but that's on the current deskside machine in the study. My server isn't powerful enough to run mythtv, though I've considered running the master backend and MySQL server on it, but no capture card.
Nor have I done anything with streaming audio, though I have on occasion looked at it. My server, being rather anemic, doesn't have much disk space - enought for $HOME, email, and link-based backups, but not for mass media.
Oh yes, these systems all run Gentoo. You spoke of getting this to run in several weeks. That DOES leave Gentoo in the running, perhaps as a decent contender. What you lose in compile time you more than make up for in generally being able to "emerge, then use". Building binaries on the machine where they're to be used generally gives fewer problems at run time.
I haven't given a direct answer to your question, just a few points for thought. I really suggest you first "define the mission" carefully before starting to build a system for it.
To Phred14 and Frank:
Thanks for your comments. You are right Frank I totally don't understand Gentoo or any other Linux distro. I've also found that there are likely hundreds of them. Debian I have had some passing experience with 2 years ago. My goal is to learn more about Linux in the process.
I'm also willing to withstand some ridicule for being a Windows worm or whatever. The Challenge to Linux users might be to offer suggestions as to what has worked for them. The Challenge for me is to digest what I learn and find a combination of applications and a distro that will work properly with both Windows and Linux clients with as little bloat as possible. Then I'd like to document what I did, Even as I am doing it and help others like me that are interested in Linux to help build their own home systems out,
I have received some criticism before for trying to run all of this on a Celeron but I have seen advertising for C3 and Geode equipped systems and a HP dual core Pentium running at 3GHz on the other end of the spectrum. All of the systems were touted as Windows Home Servers. So its possible that a Celeron might be a happy medium. I have a couple of P3 single board computers and a couple of P2's that might also serve in this role.
I am sure that if I was going to support solely Windows boxes or just Linux boxes, a standard Linux distro would work and using the Windows Home Server would work for Windows and either could easily be built in short order.
As to defining the feature set of the Linux-based Home Server. I'm trying to limit it to what I have seen specified for the Windows version at least at the start, Generally the idea is to generate an install that will work with Windows and Linux desktops in a workgroup.
That also presumes that a Windows user will be able to install it as well. My test subject is (at least at first) is going to be my 19 year old son that has been able to install Windows applications since he was 8. That basically means he was able to click on the selection in a popup window and run the MSI package for the install. So he's more typical than me. He's managed to add all sorts of IM clients and file sharing stuff to his win2K box to the point that I've had to clean it out two or three times in the last 2 years.
The goal, I suppose, is to create an install group of packages that can be run from either a command line or a browser window. The ideal goal would be to create a booting CDROM that can run the install without requiring a previous install Linux system. The install will require a minimal set of responses from the installer. That I am beginning to appreciate will require a level of automation to interrogate the home network enough to determine what network model would be appropriate for the server to mesh with. I am presuming that the typical windows user is not going to be able to determine what his IP is or netmask or whatever.
Feature set (at this point):
1) File server of multi-media files. NFS for Linux clients, Samba shares for Windows clients. Web browser for user interface on both sorts of clients for applications that do not support remote connections.
2) Microsoft touts the automatic backup capability of all of its clients onto its Home Server. That will require some hacking VB Scripts into forms that the Linux server can feed into the Windows boxes and have properly authenticate as if it was coming from an authorised admin user on a Windows computer. Samba looks to be the avenue to support this activity.
3) Cron and MythTV for time shifting television programmes. Not a MS spec but one I want to add. I have read that add-ons to the server with a plug-in tuner card will support this on the MS Home Server.
4) The ability to operate the Linux Home Server as a proxy. The idea here would be to allow the LHS to connect to sites, download media and then either stream it to the home network client or modify it for streaming to the client in a form that the client can playback. Obviously (or not) it would also allow storing the media files on the server.
The Home Server does not use the typical Windows (really MSDOS) style of "drive lettering". All of the drives installed on the Home Server look like one big storage bin. That will require an installation on the Windows clients to map it to a local drive letter for client initiated actions for either record and store or for playback activities. This model of storage is already in play on Linux systems.
Installing a Samba client on the Linux box would likely be all that's required for Linux interaction with Windows Home Server.
5) Non-duplication of files using single instance storage is a touted feature of the MS Home Server. That feature is using one of the Server 2003 R2 features. I have read that in Linux non-duplicate files can be implemented using a script. File links look to be the mechanism for that. That might make file permissions difficult to maintain, I'm not sure of that. I wonder if a smaller clip of a file can also be handled using this mechanism.
6) Bare-metal recovery of connected clients is a feature of the server. A combination of scripted PNX transactions and a startup CD sounds like the way to do that.
7) The Home Server from MS does not include DNS, DHCP etc but it should. For the initial tests I will not include those servers on the Linux version but that is such a reasonable thing to do it makes no sense not to. I currently run DHCP and DNS on the SUSE 10.1 system that I will leave running while the other SUSE system's hard drive goes into storage and it becomes my test rig.
8) Print serving through Samba for Windows clients, direct CUPS for Linux clients. I'm not going to worry about one-click service for printer driver installs on Windows clients. That would be difficult to arrange since it uses not only SMB but also Windows RPC and BITS services to do it.
As mentioned by anonymous, look at Geexbox .. its a great starting point for a home server. Geexbox tries has a tiny footprint - to be run off a usb stick or mobile phone's FS, but has formidable multimedia & TV capabilities - mostly thanks to mplayer. Expanding the footprint to incorporate firewalls, filesharing etc. should not be too challenging. (GeexBox already has quite a lot of LAN P2P features).
For the backup solution you may want to consider using Bacula.
http://www.bacula.org/dev-manual/What_is_Bacula.html
It support a lot of operating systems including Windows with the exception of Vista.
Full, Differential, or Incremental backup.
http://www.bacula.org/dev-manual/Disast_Recove_Using_Bacula.html#RescueChapter
This link gives info on how to do a disaster recovery & including bare metal recovery of linux, bsd, solaris or win2k/XP systems. The procedure is not automatic however as there are commands to enter on the system to recover, but maybe a script could make it so...
So it looks like a good solution, eventough they warn that it is not painless to set up.
I've download Geexbox and burned a CD and played it. Not sure if its a server answer or not.
I saw the Bacula package on The Ubuntu 7.04 Desktop distro, will be trying it out probably today or tomorrow. I suppose Bacula is a play on Dracula. Its installing right now on the Ubuntu desktop install. I looked at the Server version and that will be what I'll probably use if Ubuntu is the distro I use. But until then its nice to have a desktop so I can read the doc files easier.
I don't consider Vista or Winsta as I like to call it, a viable OS. Its too fat, too demanding of hardware and the pocketbook and not a proven entity. I don't understand why Microsoft thinks that flushing the user interface down the toilet with each new OS is such a wonderful idea. It aggravates the hell out of me that just todo the same things I was doing 5 minutes ago on an XP box I now have to learn a new way to do the exact same things on a Winsta box? How stupid is that?
BTW Winsta ryhmes with gangsta' since that's the way I feel about MS.
This comment has been deleted at the users request
I've been waiting for MS to flesh out their WHS description and found nothing newer than this URL.
http://www.microsoft.com/windows/products/winfamily/windowshomeserver/features.mspx
These below are the specific features I was able to discern from the advertising on the URL citation listed above. There is also a mechanism that allows 3rd party applications to operate within the server framework, I'm going to assume plug-ins.
Digital Media Shares
media and other files served throughout your home and through a website IP
Remote Access to Your PCs
WHS lets you access your PCs and media from a Web-connected PC
Protect Your Data
automatic backups and full system restore. Smaller restores of individual files and folders (incremental daily backup) of accidentally deleted material.
PC Health Monitoring
Manage a family of PCs with WHS monitoring . Make sure virus definitions are up-to-date and each PC is running the latest updates.
Users rights and access management
Users can be grouped and allowed specific access to shares and PC access
Firewall (assumed)
Microsoft can [u]NOT[/u] be so stupid as to not to build in a stout 2 way, state-full watching firewall. I will not be that stupid in any case with the Linux contender I build.
Web-Browser management and access
This is also an implied feature.
I've decided to run Ubuntu desktop 7.04 to do further testing of Linux applications, because its currently loaded on my test machine. I'm not convinced it makes a hill of beans difference what I run at this point.
(see update below VVVVVVVV)
I plan to at least attempt to pick matching applications for use on a Linux Server based on what services have been publicized as on the Windows Home Server. I don't have a Beta of the WHServer to test unfortunately so I am going to have to go by what Microsoft decides to release. Actually the feature set is amazingly sparse once you get past the fact that they touted the same damn things 2 or 3 different ways. An excellent example of advertising fluff, say a lot about practically nothing at all.
Once I've gotten my feature set Linux replacement applications chosen, I'm planning on using a mostly non-scientific method by selecting particular applications for each feature item purely by the unique number of positive hits they get when typed into a Google search window for 100 responses on one page.
Another un-scientific decision point, if I find too many negative issues based on the hit count, that application, or at the very least, that version of it gets dumped. Once all the feature points are covered, a test of the applications one by one will be made.
Once I learn how to manage them properly, a set of feature covering applications will then be installed on the no-head version of Ubuntu Server 7.04. The entire set of applications and the operating system will be configured and tested.
The step after that will be to add what applications and services that should be added to make the Linux Server a much more attractive system based solely on the feature set and we're now quickly beyond Windows Home Server. In the mean time I'll be updating in a random event basis on the blogs but certainly on average once a day. After I feel like there is a reasonable coherency to the “mashup”, I'll put it together in a secondary distro and probably hand it off to the Ubuntu group assuming its still running on Ubuntu at that point.
As far as programming, I hope to do as little as possible. I will however write/build/test scripts to automatic the entire installation process to the maximum point I can. Until I can talk to the box and make it do what I want it to in a human language, its something I have to do but I'd really rather not!
Update to the above
After struggling somewhat with the Ubuntu 7.04 desktop I decided to take a new tack suggested to me more than once. Try Debian, So I installed the inst-net iso on my test box, had Samba up and running in about 2 minutes and I'm already much happier. Enough so that I probably will clone the drive at least once so that in case I manage to hose it I can start again without delay.
Response time of the desktop is snappier with the Debian install on it compared to the Ubuntu, not sure why but it is appreciable.
Otherwise the plan is to continue in the same method as listed above.
I do know that MythTV will be the last application I will test since it seems to take over the system as soon as you install it. I probably will dedicate a single machine to become my new "TV server" if I do decide to make a MythTV system run.
Speaking as an interested lurker...how is the project going? Any updates?
Update 6-13-2007
Since I don't make money working on Linux, I've gotten into a bit of a time crunch and so I apologize for my update delays.
Bacula has such an attractive feature set I think I'm going to break my own word and build it. I found a couple of Help pages that will certainly make the process a lot more painless.
HOWEVER, this isn't the method the typical USER is going to be able to handle. I'm not going to expect everybody to whip out their C, or C++ compiler and build an application for their own installation!
If I can believe what I read on the Windows Home Server website, RC1 has been released and I didn't make the cut for the release. So I'll be building the Linux Home Server without a Microsoft product to compare it to. I'll check with the guys at work and see if any of them got into the WHS RC1 or beta program.
In any case the comparison really isn't that much of an issue now since it seems as if it will a lot more interesting to combine Linux applications together instead.
More to come!
Strange thing you say about Bacula... i haven't seen any information about needing C/C++ knowledge to operate it. It's only about modifying some config files and running commands... Shouldn't be too difficult if you only plan to backup a couple of machines on one server.
So what's the status now, what solutions have you tried yet, now that almost 3 weeks have passed?
Lurkers welcome Karen.
ShakaZ-- I think I probably busted the chain here I posted another blog by mistake at this Heading
Software Selects for the Linux Home Server 6-14-2007
I also added a comment. Turns out my "whining" about WHS might have gotten me an invite to download RC1 of WHS, more to come."
Slaving over a hot DVD/CD Burner making WHS-Stew
Well it looks as if everything for Windows Home Server RC1 got burned onto DVD-R and CD-R's OK. Discs are all readable.
Then I realized that my Celeron test machine doesn't have a DVD reader! Now I have a bit of a reason to delay the inevitable for a while longer.
Yes that's right, the WHS RC1 was a DOWNLOAD! (Gee I wonder where they got that idea?) I think that this is a first for MS. It is NOT available on disks for order, at least that I was able to determine. The ISO for the install on the server is a DVD image a little over 1.1GB. No there weren't options for a couple of CD-ISOs instead.
A second ISO of about 296 MB is a Windows bare metal restore disk that probably loads Windows PE (Pre-install Environment, fancy name for No-GUI Windows) and talks sweet to the established WHS and restores the current backup image from there.
Finally the third ISO was a 37 MB image of a "Connector" installation disk that you would use to set the Windows' ONLY client systems up to connect to the local WHS. This method has been often used in the past for installing Exchange server clients, Proxy Server, ISA clients, and Active Directory installs on NT 4.0 workstations and Win9X clients of an Active Directory Domain. You'll notice that most of the products I mentioned are all before "push-technologies" that Microsoft has produced in the past.
Remote Terminal Services would be an ideal means to make this work. But I'm willing to bet that the default security settings of SP2 and the hundreds of quick fixes since then probably won't allow MS' Terminal Service to work at installing the software.
Trying to keep it all in the family, I downloaded it to my Windows XP SP2+ system. I used Nero and followed the advice given me on the website specifically about Nero so I shouldn't have any issues there. The test drives I will be using all comply with the recommended sizes and with the 512 MB RAM I'm ok there as well.
I haven't tried installing the ISOs yet and I will wait until I'm able to do it with a DVD burner installed in the Celeron test machine. I'm very loathe to give up my Little Debby Linux box though (currently running on the same Celeron) so I think I will resurrect my old 1.8 GHz P4 and do an Debian 4.0 install on that one so I can keep learning more about Debian even as I'm testing the 2 Home Server installs.
Setting up the Debian client will give me 2 different Linux clients (the other is SUSE 10.2) to connect to both types of Home Servers and that might be instructive as well. If I see any significant differences in performance I might try using Ubuntu and Fedora since I have current ISO releases of those as well.
I do have another XP Pro client and a Win2K Pro client (my kids machines) so there will be plenty of clients to beat on both HServers.
Having dual-booted Linux and Windows using GRUB before, I do not see much of an issue of doing that on my Celeron test box for the 2 Home Server installs. That should allow me to try the same functions almost back-to-back after rebooting from one server to the other. It will also give me a good idea what will happen when the client suddenly experiences the server going off-line!
More to come.
This comment has been deleted at the users request
I posted the Linux line-up for my Linux Home Server at this URL
http://community.zdnet.co.uk/blog/0,1000000567,2000440676b,00.htm
and I'll be continuing my Home Server "quest" there.
