Friday Laugh: Going Back To That Other OS

SJVN found a great story for the week, a guy who had his GNU/Linux server fail.

What was the problem? An update to GNOME failed. Apparently he was using a GUI to manage the server. That’s not wrong especially if users are going to visit the server with a GUI and he wanted to have their experience. There’s more than one way to do that of course, and there’s no need at all to have the GUI on the server. It can be on your client machine. You see, a GUI is much more complex than a server needs to be and a server is automatically more reliable if less software runs on it. It’s statistics. No matter how good your programmer there is a good probability of having X bugs per thousand lines of code. A server without a gui is millions of lines of code. A server with a gui is probably many times more buggy. The GUI failed when he updated it and he could not get the GUI to start… If he were running SSH on the server he could get in just fine.

The guy also complained that the GNU/Linux server was more work to maintain to which SJVN replied he spends about an hour a year maintaining his GNU/Linux server. I would agree with that. I did have a bit of a problem with the server that runs this blog but it was our own stupid fault, not GNU/Linux. It ran for months after our little tweak and eventually was swapped for failing hardware so I have not had to look at it for a long time. It’s just there. Where I worked last year there was no running server in the building and I set up several with no GUI, not to make work for myself but to make my life easier…

GNU/Linux servers are so reliable that I recommend every user should have one. If there is no server hardware on your LAN, you can run a service on your client. There’s nothing like connecting to a web application with no network lag. I had to warn students lest they fall of their chairs. Just install Apache2 or Boa or Nginx and you can serve static files instantly. Add PHP or perl and you can do server-side scripts. A database is real fun. You can make servers very complex or have hundreds of them but one server should never be a chore but a blessing.

see SJVN Linux servers work just fine

About Robert Pogson

I am a retired teacher in Canada. I taught in the subject areas where I have worked for almost forty years: maths, physics, chemistry and computers. I love hunting, fishing, picking berries and mushrooms, too.
This entry was posted in technology. Bookmark the permalink.

46 Responses to Friday Laugh: Going Back To That Other OS

  1. oiaohm says:

    Robert Pogson I am young enough that I never had to work with punched cards unless it was data recovery. At this point I was wanting to kill someone. The machine auto feeder of cards had been broken so I had to manually feed each one in.

    Drive with data on failed. Tape failed and bugger there is the hold cards do I really have to re enter those. Ok you have two options go back to the manual paper copies of data or or make the punch cards work and we will do from when the punch cards ended.

    Yes I have very bitter memory of data-bases on punch cards.

    So you have done the evil job of making those punch card databases from nothing. I really do fell sorry for you. I hate them enough when they are the only records left.

    Also just to be evil the records I was dealing with long into the dumb terminal frame were they printing the records out to punch cards for long term storage. Since punch cards have a very long storage life compared to tape and harddrives.

    Luckly the newer stuff. Placed at start of card a index key number punched into card then data. So even if you put the cards into the machine in the wrong order the data was fine. The were even smart enough to make the last line end of card half zeroed and half alternate. So you could not put the card in upside down or mirrored and stuff up the input because the last line on the card being first would show the error same with being around the wrong way.

    So I did not have the evil of having to keep the cards in the right order. But if one was missing had to find it.

    Yes you could in the system I was dealing with remake a damaged card that had been read in wrong due to the damage and it would auto replace the same index record data. So each card was very much like a row in a conventional database today.

    Most likely I saw the punch cards a little father down the line of development to the databases of today.

    Yes the latter ones cured the problem of the user error with cards. Yes the language word key/keys in databases evolves from punch cards.

    But I did learn some good ideas for data storage from them.

    Yes this is why I get annoyed when people mix up the term database and spreadsheet. Both have a very interesting history and a lot of humans who suffered making them what they are today and we are not finished messing around with databases yet.

    Also both the people who invented particular things are lost to history. The did not class it as important enough invention to record there name to it. Yes IT world has its equals to the inventors of the wheel that we don’t know who they are be but depend on what they invented every day of the week.

    I had some short programs I had found in the process. Issue “sequence number” This coder was kinda evil. 10 cards no sequence number you had to be smart enough to read the code and put them in the right order so it worked. One turned to be a basic game of hangman. Sign of program was no sequence number. So it was at least simple to sort them from the data cards.

  2. I am old enough that I used punched cards for about a decade. For a few decades after I still had images of those punched cards, thousands of them, on magnetic tape. We did have quality control with punched cards. You young whipper-snappers may not realize that punched cards often had sequence numbers for sorting and keeping in sequence and keypunching was a skilled art. The standard operating procedure when you were serious about getting it right was to have multiple people enter the data and compare the results. Difference == visual check or re-entry. My first paid job as a student at University of Manitoba was to punch in FORTRAN programmes and data for a statistician. They had hundreds of thousands of punched cards in their library. Fortunately a few years later the punched cards were phased out and we used datalinks and “dumb terminals”. I’ve been there and done that.

  3. oiaohm says:

    Ivan I had not noticed that in my chroot I had pulled in a few old packages. “No distribution ships the version of boost the package depends on.” Debian still ships the old versions of packages. Has means to pull from a particular time point for a reason.

    Debian you can pull in old packages if you requirement them. Also in fact it runs with newer versions of boost as well as long as you did install the backwards compatibility packages.

    Robert Pogson and ch. Both are you failing to see the difference.

    “I can remember when shelves filled with punched cards were important databases for statisticians. ”

    If you ever handled those you would find there is a quality control system.

    There are database and data collections. All databases are data collections. But not all data collections are databases.

    Now a person using punch cards can choose to disregard pagenumbering and checksums.

    A data structure does not have to be a database either. Its in the term data collection.

    All the historic computer databases have some form of quality control. This is something all history spreedsheets don’t contain.

    Spreedsheets have means todo maths on the stored data. This is a requirement to be a spreadsheet.

    Databases only contain data in data areas. Where spreedsheets store formulas that are calculated. Formulas that on a real world version would be performed as part of procedure and that value directly entered into a cell.

    Databases and Spreadsheets are different.

    Yes one of the creators of databases where statisticians before computers. They collect forms box them that is there database. They had to run quality control on the data to make sure none was missing.

    What a database is very truly exact. What a spreedsheet is very truly exact.

    The problem with calling one the wrong one is yes spreedsheets can be used as list. But they don’t have options of unique keys. Unique keys this like the number of records in a box. So if you remove a record a mark is left behind until you add a new record in.

    Anyone who as deleted a row in spreadsheet what happens the row under takes it row number.

    Next is a traditional limitation spreadsheet you might half way down the sheet start a new set of coloums and start entering a different set of data.

    Database this is a different contains a different set of data is a different table/box/shelf/what every storage grouping you are using.

    Database and Spreedsheets have two different rules.

    A spreedsheet can be made partly behave like a database. But when you start talking auto maths spreedsheet a database does not do this at all.

    Calling a spreadsheet a data source or a data collection is valid. Calling it a database is making a mistake it don’t truly operate by the rules of a database.

    Both Databases and Spreadsheets are data collections. Basically people are using the wrong generic term because everyone forgets that data collection is a generic term.

    statisticians are where views and other things in databases come from.

    Spreedsheet are from book keepers in history.

    History has been forgot here. Fragments are remembered.

  4. Ivan wrote, “No distribution ships the version of boost the package depends on”.

    Ask Amazon why that is. If it’s an old version, there’s no problem running an old version in a chroot or virtual machine. If it’s a new version, one can install the new version. One way or another a person could install it.

  5. Ivan says:

    “Ivan! You are not the Lord and Master of anyone here.”

    Never claimed to be, Bob. Maybe you should take a breather before your heart explodes.

    “Do the installations yourself and give us a full report if you have the time, please.”

    I have, many times, that’s why I know that Mr. Ham is lying through his teeth about installing it. No distribution ships the version of boost the package depends on.

  6. ch says:

    “around 100K recipes”

    Sorry, but that’s not a “recipe collection”, that’s an encyclopedia 😉

    “Excel spreadsheets that probably should be in a database”

    Agree. Excel is great until a point – but then you should know to move to a better tool. Mind you, I’m not saying that one should put everything into Excel – definitivly and emphatically not ! – but just that Excel is one of the available tools. When MS asked users what they did with Excel, they were surprised themselves how many people used it for “lists of things”, and so the MS developers put in some basic DB tools like Autofilter etc. Another rational use for Excel is to connect to a bigger DB, suck in data and then process it – statistical functions or calculations above those that SQL allows. For this, a million possible rows can come in handy. Working on SAP data that way is quite popular in accounting, I’m told. (As far as anything connected to SAP can be called “popular” 🙁

  7. I have a database on my system of around 100K recipes. It’s just silly to think of scrolling though the instructions looking for something. A proper index is needed. Probably for even 1K recipes that would be true. With a relational database, I could index millions of recipes if I had them. A spreadsheet with a million rows or columns is an admission of failure to adapt the right tools for the job. I have read here that some businesses are locked in to M$ by their Excel spreadsheets that probably should be in a database that could be easily ported to any platform. Mixing the application with the data is a failure adopt a rational approach, modular and efficient. That may have seemed the right way to go in the old days when PCs were novel but now it is trivial to implement a database and trivial to implement an application to access a database in a standard way that there is no need for a special application to do both at once.

  8. I have to disagree, oiaohm. In computer science there are many types of data-structures. All may be databases. I can remember when shelves filled with punched cards were important databases for statisticians. They would collect the data from somewhere in some way and analyze it for years looking for subtle correlations. The technology may change but the features keep covering the same things, the abilities to create, find, change and distribute data. A database may be a handful of discrete variables, various types of lists, arrays and records in one place or distributed over many sites all over the planet. For a small enough collection of variables, a spreadsheet may indeed serve well as a database but it’s not as fast or flexible as many relational database systems. That’s why Oracle exists. Fortunately I have MySQL and Postgresql as alternatives because I don’t make enough money from databases to pay Oracle’s tax.

  9. oiaohm says:

    I missed a minor point availability is extract ability and return ability.

    Extract for use of course I was writing on the presume that you would include that it also has to be returned possibly altered/damaged/improved. This is where a spreadsheet falls apart. Start trying to remove and return a row.

  10. oiaohm says:

    Ch please read the full def of a database
    “However, not every collection of data is a database; the term database implies that the data is managed to some level of quality (measured in terms of accuracy, availability, usability, and resilience)”

    Hosts file is in fact not a database but a list. Does not have a requirement to be organized or be correct or be resilience or provide enhanced usability.

    Yes a excel spreedsheet is a collection it also fails the tests if it is a database.

    The most basic real world database is a card file box. This is the problem. First electronic forms of a database was making programs to replicate the humble card file.

    Both spreadsheets and databases come to life from real world items that are completely different. Spreadsheets exist in physical form that is very simple to spot because they are directly called spreadsheets.

    Databases in the real world are in fact many different things these include card files, Libraries ie the structured layout of the books so the library is a database in itself so you can tell if a book is missing or find the book you are looking for, organised specimen storage methods and the list of real physical forms of databases goes on and on and on.
    This is the key provide of any database system
    “quality (measured in terms of accuracy, availability, usability, and resilience)”

    The storage must be able to provide options for checking accuracy of storage. Optional numbering so if a item is missing it spoted is one such option.

    The storage must provide availability. In the sense that a record must be extractable for use. Book,specimin,card, database record are all removable items for use so meeting the requirement.

    Usability is the most subjective test.

    resilience simple take every book,specimin,card, database record out there storages following the rules of each of all those databases could you put them back in the right place with new items. If so it has resilience.

    How many does a host file or a spreadsheet fail to get over. Many of those things a spreadsheet is not built todo.

    Yes I know there is a disconnect in most books you see out there. The early history of how databases in the computer world came into existence has been lost. So leading to the term not being very well understood so it being mixed up with just being a collection.

    The electronic form of both is really a continuation of what happened in the real world. Once you understand this what is a database is very cut and dried.

    Computers processing lists like host files existed before databases.

  11. Ch says:

    “Ch simple fact excel is not a database.”

    I explicitely wanted to make the point that yes, indeed, it is. The definition of database is pretty basic and therefore far-ranging: “A database is an organized collection of data for one or more purposes” (WP). So even a structured text file (think of “hosts”) is actually a database.

  12. oiaohm says:

    Ch simple fact excel is not a database.

    There are many issues you bring by using excel instead of proper database structs.

    My biggest annoyance is my favourite database front end kexi is not free for Windows. Really for setting up CD collections or recipes with forms and getting data entry running even distributed between all computers is fast and simple with it.

  13. Ch says:

    ““easy” doesn’t scale when you end up with hundreds of columns”

    Right, but how many columns do you need for:

    “the CD collection grows, the recipe collection grows”

  14. oiaohm says:

    Ivan In fact yes I have installed them on all the listed distributions bar OpenSuse and the Suse line and many not listed.

    Fedora package I have used on redhat enterprise and scientific Linux without issue.

    Debian package used on debain and mint debian.

    Ubuntu package used on Ubuntu and many of its relations.

    Basically you are the one being a troll would I bet. Reason you have not stated once exactly how it fails for you. Ivan.

    Really please speak up maybe you found a bug I missed. I will give that amazon version numbers are deceptive the and latter got loss somewhere even that they work perfectly on latter. Like the debian I used is debain 6.0 squeeze not Debian 5 as the package list.

    Amazon is in fact one place I do by music from. All my laptops and desktops are Linux machines of some form.

    I did think of one problem you might have run into Ivan 64 bit Linux install refusing to take 32 bit package. But I chroot around that.

    Yes on a 64 bit bit Linux system I do maintain 32 bit chroot for the few odd ball annoying applications like amazon downloader. Its under a 1G foot print for lazy installing of these odd ball apps. Large harddrives I really don’t care about the 1 G foot print.

    Did you try to install the 32 bit packages the hard way?

    chroot method.
    http://www.debian-administration.org/articles/356
    Disc effective method. That I find frustrating.
    http://www.debian-administration.org/articles/531

    Please not the chroot method is basically distribution neutral. It how you solve bugger this application will not work with newer version of distribution as well.

    debian-administration.org is a great site for all those annoying debian problems. Chroot tech applies to all distributions. Minor differences.

  15. ch wrote, “Hiding things is hideously easy in Excel, actually: Right-click coloumn head -> hide. And there’s Autofilter.”

    “easy” doesn’t scale when you end up with hundreds of columns to hide one minute and another set to hide the next. It is far easier to display the ones you want with SQL. It was invented for a reason. The right tool for the job…

  16. Ivan! You are not the Lord and Master of anyone here. Do the installations yourself and give us a full report if you have the time, please.

  17. Ivan says:

    “Amazon’s mp3 downloader works on debian 5 and up. Fedora 11 and up, Ubuntu 9.04 and up, OpenSuse 11.1 and up. Natively.”

    Congratulations, you can read. Now actually try to install and use it on the listed distributions, Mr. Ham.

  18. ch says:

    A tool should fit the purpose, and for some purposes Excel is all the database you need, e.g. the birthday list of your department. Hiding things is hideously easy in Excel, actually: Right-click coloumn head -> hide. And there’s Autofilter.
    Of course, there are limits, after a certain point one should use a “real” database, but I just wanted to point out that actually a lot of databases still reside in Excel, and sometimes that’s actually justified. On the other hand, how many people do you know who speak SQL fluently ? That’s where dBase and now Access come in: Unless you have 15 PCs in your house, the Jet DB does the job quite fine, and if you really need more, you could use Access to access data in a SQL Server Express DB.

    In a corporate network, SQL Server has the big advantage of AD integration. However, for stuff like SAP (AFAIK the DB doesn’t lend itself well to clustering) my company prefers a big Sparc-server with Oracle (and Solaris).

  19. If you were one of my students you would know that spreadsheets don’t scale nearly well enough for many purposes. Occasionally a visitor here will comment that M$’s spreadsheet is better than OpenOffice.org or LibreOffice because it handles million-line spreadsheets for efficiently… That’s long past the point where a relational database should be used. There are many reasons for this. Big ones are indices and scrolling. Being able to find something in a huge spreadsheet is problematic. It all has to fit in RAM or is hopelessly slow. A proper database like MySQL or Postgresql do not assume the database will fit in RAM and do the right thing. Then there is the ability to quickly display some columns and not others. With a spreadsheet you can hide things but it is cumbersome. With a database Select is quite natural. When it comes to sharing a database on a web interface, a spreadsheet really is limited whereas many web apps use a relational database quite effectively.

  20. Nonsense. Price/performance is key. Waste can be defined as too high a price for too little performance.

  21. ch says:

    “Everyone thinks they don’t need a database at first but the CD collection grows,”

    Probably the most-used database is actually Excel, but for more serious stuff you might try Access – a database server will rarely be needed in a household.

    “As it was we took the lowest bundle which may not have been optimal.”

    But, but, but … you always told us that cheaper is better and everything else is WASTE !

  22. oiaohm says:

    oldman ?? What are you talking about.

    There is closed source desktop software for Linux. SoftMaker Office 2008 and others.

    I guess you mean No Microsoft software so it cannot have market. Oldman.

  23. oldman says:

    “Yes lot of the difference is being sorted out in recent cooperative work between distributions.”

    No closed source commercial desktop software, no major audience.

  24. oiaohm says:

    Dr Loser

    “Ever heard of virtual machines?”

    Yes I have cgroups and chroots both have lower overheads. Reason no emulation of hardware. There is only 1 kernel in the mix as well. So you can use socks to mysql and postgresql and other things in different system roots with cgroups that you cannot use with virtual machines so again avoiding overhead by not duplicating up services.

    Yes cgroups and chroots are part of a form of virtualisation called http://virt.kernelnewbies.org/LXC

    “I don’t know about you, but I don’t live in a household with dozens of other people continuously streaming exactly the same thing.”

    You really are not looking too hard. Squid setup right these days can cache youtube videos and many other video site videos. Mostly because it can go by the sites unique id of the video. So simulations streaming is not required to get the reduction.

    http://eu.squid-cache.org/Squid-2.7 I guess you were not up to what Squid could do these days with fairly simple rules Dr Loser.

    Of course modern day Squid can be caching the Windows update downloads as well. So giving you the advantage of saving on downloads as running a WSUS without having to configure the clients to update from WSUS.

    So yes with squid-2.7 that part video download from youtube that broke than when you reload you have to download completely again does not happen that way. Since the resume can now start from a different source server.

    Of course the MS proxy server is still a useless bit of works.

    Yes a wall wart can run a squid-2.7 caching youtube windows updates and other things giving a nice performance boost and making the case you want to show a video to someone else in the house hold lazy. Just give them the link and they get it from proxy instead of internet.

    Ivan are you just asking a stupid question? I guess you are running on Windows so when you visit the Amazon’s mp3 downloader you don’t see the Linux download page.

    Amazon’s mp3 downloader works on debian 5 and up. Fedora 11 and up, Ubuntu 9.04 and up, OpenSuse 11.1 and up. Natively.

    Distributions based off those it also works on. About 80 percent of all Linux distributions. Ivan. Of course Mint make sure you check what base its on. One is debian based one is Ubuntu based and there is a difference.

    Fedora 11 package covers Scientific Linux and Redhat Linux most likely centos as well.

    Of course using chroot you can make that run on them all. Ivan question is how much effort you are willing todo so you can run it. Enough effort you can run it on them all. There is really not that much to the Amazon’s Mp3 downloader.

    Please define effort level Ivan.

  25. Ivan says:

    And which of the distributions lets me use Amazon’s mp3 downloader?

  26. A fellow would not go wrong choosing any popular distro such as Ubuntu, Debian GNU/Linux, Fedora, RedHat etc. These have been around for years and will continue supplying end-users with important quality control and maintenance services. Any of these should be able to keep a server tuned up very nicely indefinitely. I recommend Debian GNU/Linux because it has the APT package manager, a gentleman’s updating pace,

  27. I would agree, if it never sees the web. Then you don’t need updates or re-re-reboots. If the machine is ten years old, it probably takes a minute to reboot every month. That’s 12 minutes right there and you haven’t accomplished anything. Further, a server has actual useful data which a person may wish to backup and ports open. That requires more maintenance than a client machine.

  28. Dr Loser says:

    @Robert:

    Oh, and that huge expense in time and licences?

    I’m sitting on a grotty ten year old XP machine, and although I’d appreciate the incremental benefits of moving up to Windows 7, it’ll do for now. As with your mythical one hour per year Linux server cost, my little XP darling costs me more in cleaning out the fluff from the fans than it does for software.

    And more importantly, I don’t even notice it.

    Let me put it very bluntly: the “cost” of administering this tatty ten year old PC is, for all intents and purposes, zero.

    The up-front cost was to create an Administrator account (which I hardly ever use, except for Spotify, which is basically software junk with valuable stuff behind it). That was, what, five minutes?

    Beats your stinkin’ one hour on the server.

  29. Dr Loser says:

    @Robert:

    Sorry: I phrased the question the wrong way. My fault.

    What useful function does the current situation, re distros, perform?

    For somebody (not me, not you) to look for The Linux distro, they would practically have to navigate the entire historical tree of distros.

    And then they’d have to deal with the fact that some of them are flavour of the month, but only this month, and some of them are up-to-date with upstream on this but not that, and so on.

    It’s a valid question.

    And it goes to the heart of the problem that the bloke you are blogging against was describing.

  30. Dr Loser says:

    @Oiaohm:

    “With more than 1 device downloading internet content in a house today a simple proxy server can reduce internet bills and increase over all performance.”

    It’s about the only intelligible thing you wrote, and it’s nonsense.

    A proxy server is an overhead.

    The only way that a proxy server could possibly save on either bandwidth consumption or on latency (“internet bills and … performance”) would be through caching.

    I don’t know about you, but I don’t live in a household with dozens of other people continuously streaming exactly the same thing.

  31. A person could choose to assemble dozens of applications from all over to run on your PC or you could let experts do it. Thousands of packages in the Debian GNU/Linux repository are ready to run on Debian systems and there’s very little the user has to know about their inner workings. The advantage is that I can update my OS and my applications from a single source. That saves me most of the work of managing a PC. Further, because the repository is Free Software, I don’t have to agree to dozens of licences as I install or start the applications. The net result is a huge savings in labour and expense.

  32. Dr Loser says:

    @Robert:

    “As are Fedora and CentOS. Developers working for the different distros. The end user should not have to muck with porting applications. That this guy did was one of his mistakes. There is a reason distros exist.”

    That’s an important question right there. What is the reason for distros to exist?

    Don’t explain it to me; let’s assume I either know or am bigoted (I’m accepting both).

    How would you explain this, say, to your students?

  33. Dr Loser says:

    @oiaohm:

    Ever heard of virtual machines?

  34. oiaohm says:

    Robert Pogson its also the reason why cgroups and chroots exist as well. If you need something from a different distribution you can install it contained.

  35. As are Fedora and CentOS. Developers working for the different distros. The end user should not have to muck with porting applications. That this guy did was one of his mistakes. There is a reason distros exist.

  36. Everyone thinks they don’t need a database at first but the CD collection grows, the recipe collection grows and pretty soon one has 20 databases in use all with a different user interface. 20 years ago my wife and I built our house using a DOS spreadsheet. We should have used a database. It could have saved us $thousands on the project by allowing us to sort out all the competitive bids more easily. As it was we took the lowest bundle which may not have been optimal. Databases are fun because they scale far past spreadsheets. I know some like scrolling over hundreds of rows/columns but I don’t consider that fun at all.

    Every family has tons of stuff/information collected. It makes sense to have them all in one place in a database.

  37. There are some excellent web apps for databases and using a web browser for the GUI makes sense. Everyone knows the web browser so there’s less training/practice required to use the database. Even for the recipes it makes sense to use a web app. One only needs to install once and everyone on the LAN can access it. Less work.

  38. oiaohm says:

    Dr Loser Unity is not Gnome 3.x or Gnome. KDE is not gnome either. Neither is Lxde, xfce and a long list of other options.

    “(You’d need Gnome, of course. Apparently upgrading to Gnome 3.x is not recommended.)”
    This basically proves you are complete moron on the topic because gnome is only an optional part. So I will start off with basic reasons.

    Owncloud.org should be your first stop. This provides a nice little server you can sync your ipad iphone and android devices against without broad casting your private information to the world. Also provides a nice safe place to download emails and other things to.

    With more than 1 device downloading internet content in a house today a simple proxy server can reduce internet bills and increase over all performance. Local server can serve up files many times faster than the internet in most countries. squid of some for is quite suitable for this usage.

    There are advances coming allowing GTK and QT based applications to be provided to ipad and android pads by html5 including rdp to windows machines as required. http://www.ulteo.com Yes commercial grade solutions as well go down this path.

    Media storage for the Xbox 360 and Playstation.

    Then you can be running http://Status.Net so you have your own backup of everything you have send to identi.ca and twitter so that if anything goes wrong you can upload to a new service provider no problems. There are other on-line services that can be federated locally.

    Possible backup server for you windows clients using 1 of many options you can do.

    Target audience is basically everyone who has more than 1 computer sharing a single internet connection. Since there is an advantage there from shared caching.

    Another target audiences who what want to backup there private data to a server they control. So that if anything happens to the remote server the data is not lost.

    Another target audiences is those who want to do more with there ipads and android tablets.

    As you can see very long list here. Of course sizes of these servers could be anything from something small enough just to plug into a wall socket as a wall wart. To something the size of a normal PC.

    Yes a wall wart containing an duel core arm with a 100G hd would make a very good proxy server for 10 to 20 machines. Should be more than suitable for the home market.

    There are some distrobutions out there targeting different sections of the market.

    Even some support backing up your cloud server images. Again it depends on what you are doing.

  39. Phenom says:

    Dr Loser is right. 10 years ago I always kept a server at home, mostly for db and web services, I needed it for development. With the maturing of development tools, that’s no longer necessary. Now I only keep an external HDD for file storage, attached to the router. Technically, it is a file server, but definitely nothing more, and absolutely minimalistic.

    Hosting a web app at home? Please. That’s the stupiest thing to do – trade desktop usability for some web-hyped half-assed subst.

  40. Dr Loser says:

    OK, I’ll bite on the secondary advice. Why on earth would “everybody” want a server, Linux or otherwise?

    And what you’re describing, Robert, isn’t a server. It’s a desktop with a few daemons (Apache, etc) bolted on. Without a network, it isn’t “serving” anything.

    I would consider it to be a fruitless complication to “serve” static files to my own PC: there are plenty of HTML viewers out there that do that job for me. (You’d need Gnome, of course. Apparently upgrading to Gnome 3.x is not recommended.)

    And databases aren’t fun at all. I don’t know anybody who gets his rocks off on a database. But if you have to play around with them, you don’t actually need a LAMP stack and a welter of Perl or PHP scripts to do it; just talk directly to the database.

    Is there a target audience out there for this Server One Box idea of yours?

  41. oiaohm says:

    ch once you get past the window dressing there is not major differences between Linux distributions.

    Configuration files follow Linux Standard Base in Ubuntu CentOS and Fedora.

    Today package names between distrobutions are mostly the same name for the same application.

    We are almost to the point the only difference is package manager engine it self.

    Most people have not noticed the difference disappearing. Ubuntu is going to be a edge when systemd goes threw the distributions. Yes lot of the difference is being sorted out in recent cooperative work between distributions.

    Yes more and more different distrobutions are coming the same OS just different versions.

  42. ch says:

    “He talks about how using Ubuntu package instructions won’t work on CentOS or Fedora. Well, no, they wouldn’t. They’re different distributions, different operating systems.”

    And why exactly ? What good does that do compared to the bad of making things yet more complicated ?

    “It’s like using Windows Server 2003 instructions on Windows 2008 R2.”

    You have a much better chance of that working because they _are_ the same OS, just different versions.

  43. oiaohm says:

    JairJy Server normally equals lightweight wm if you have one.

    Gnome is feature rich what makes it large code what also makes it buggy.

    “But not on Linux, because on Linux newer software means untested and unstable software.”

    Newer can equal more stable the most critical is what it is with Linux or windows.

    Zentyal is a good example of a server setup light wm, webbase management and a light file-manager.

    Usable there are a lot of usable wm for Linux.

    People mix up feature rich. Windows 2008 still looks like Windows 2000 not like Windows XP or Windows Vista. Reason less features more stable interface.

    Yes Microsoft uses a less feature rich wm on servers. Basically its horses for courses. Gnome or KDE on a server most likely someone has stuffed up. They are both over kill for the job. So risking more problems.

  44. JairJy says:

    I really hate when someone minimizes others’ problems by saying that he/she doesn’t have those problems. It’s like “Hey, you have X problem, but I don’t have any, so, it’s your fault”.

    Some believe GUIs are needed for modern computing usage. David Gewirtz only wanted a simple GUI to make their server more user friendly. After all, GNOME is the most usable DE on Linux, it must be very stable and secure, isn’t it?

    David Gewirtz’s problem was because he wanted to update their software. Everyone knows that updating a OS makes that OS more secure and stable. But not on Linux, because on Linux newer software means untested and unstable software.

    I don’t know, maybe he is not ready for Linux. He could not understand how Linux works. He is just a dumb user who wants a User Interface and an updated software on his server.

  45. Ivan says:

    Didn’t you just tell people to shun nginx? http://mrpogson.com/2011/10/13/nginx/

Leave a Reply