You can't predict the future of share prices by using past data.
No, but you can make a fairly-accurate educated guess about the price of Company X's share price tomorrow, based on the news today. The main issue about attempting to predict the future based on history is to know enough about the effect you are attempting to predict to understand its causes, and therefore be able to accurately detect the symptoms in time to react to them appropriately.
Of course, if we're detecting revolutions, then a fascist state would simply jail or otherwise dispose of anyone who performed any action they deemed to be a precursor to fomenting rebellion, thus removing any visible precursors. Of course, doing that might be the final issue that pushes the unrest over the cusp, and starts the revolution...
If I recall correctly, one of the postulates of Foundation was that the general populace should not be aware of the existence of the prediction mechanism, or it would fail in its purpose. Also, the groups of people whose actions were being predicted was so large that an individual person's actions should have little bearing on the results.
This concept has been summed up quite well by Agent K in Men In Black: "A person is smart. People are stupid, panicky animals... and you know it."
Asimov wasn't alone. Several authors have put forth the concept (my favorite is actually "Cleology", introduced in "In the Country of the Blind" by Michael Flynn, complete with some graphs and charts of historical cyclical data that projected (at the time of the book's writing) future events that have turned out to be surprisingly accurate (at the time of this writing)).
The entire concept of studying history can be summed up in the phrase "Those who do not recall history are doomed to repeat it" - in other words, we need to know what mistakes we have made in order to avoid making them again.
Tracking cookies track. This is not news, this is anticipated and expected behavior. This has been the status quo for over a decade.
Cookies have a security feature in that they are accessible only to the websites that placed them, but advertising sites have been using tracking cookies for as long as cookies have existed, and getting around that security by placing a "bug" on third-party sites. They used to (and probably still do) implement this as a 1x1 "spacer" image the same color as the background, or simply by having an ad on the page you are viewing. When your browser requests the image/flash/javascript/whatever, the site it comes from is suddenly allowed to access their cookie.
The solution has also not changed; either don't allow cookies, or delete them constantly. Anti-scripting addons are also helpful, as are black (or whitelists) of websites to disallow (or allow) access to your system. Modifying hosts files has been a semi-successful method, as well, in that requests sent to specific named addresses can be redirected to localhost (and therefore "blocked").
I personally use NoScript and AdBlockPlus for precisely this reason (and to speed up my page loads), and I can't fathom why this information could be conceived to be news to any user with any amount of technical knowledge and a modicum of interest in their own privacy.
The paragraph in my above post that starts "As for firing up dd on the raw device" should have included "on a currently active system" for the rest of that paragraph to be valid and make sense.
I was pretty sure that dd was going to be my best friend, but we could wish for a better best friend. As near as I can tell, I ought to boot single-user off a CD and do it from there. Is that overconservative? No-way-no-how can I just far up dd on the raw device underneath a bunch of partitions, can I?
Look for dd-based partition-manipulation tools; possible search terms might be "Linux Ghost" or "disk image dd". There are a plethora of solutions out there, and you can probably find one that suits you with only a small amount of experimentation. Many are available as a bootable CD, and come with a menu system or other helpful tools. Several allow you to backup to a massive variety of targets, anything from simply backing up a partition to a gzip'd archive on a separate partition to streaming a bit-wise copy of the entire disk over the internet to a remote system you have access to (and creating an exact duplicate of the partition/drive on that target).
A LiveCD of your favorite distro or "rescue disk" or "recovery tools" can also be extremely useful; Most either already contain some sort of disk/partition tool or allow it to be installed on the RAMdrive the system boots into, assuming internet access. For example, gparted is a graphical tool that will allow you to manipulate partitions, resizing, moving, cloning, etc... and it's available on the Ubuntu LiveCD by default (last I checked).
A persistence-capable "external" (USB or otherwise) boot device can allow you to keep configuration information or notes (such as which drives/partitions you have backed up, when and to where) between boots, as well as allowing a portable system that contains your preferred set of tools. Taken a step further, you can create your own LiveCD once you have a preferred setup (assuming you're not scared to learn even more).
As for firing up dd on the raw device underneath a bunch of partitions... well, your swap will be corrupted, for obvious reasons, and any disk writes during the read process will harf your data, too, so it's not necessarily the best plan... but it's theoretically possible. At this point, I would examine your motives for imaging the disks in the first place, as a direct image may actually take more time than simply backing up your configuration and data files, and keeping a list of where each file belongs and what packages need to be installed if the server should be "hit by a bus".
Disk imaging can be a simple solution, but a reinstall can be an excellent opportunity for "cleaning out the cruft", and may actually require less time or effort than imaging a full disk, depending on the size of the data involved and where the server is located (ie, your ability to gain physical access). I've lost count of how many times I have blown a new OS onto a machine, dropped the appropriate apps/data from backup into the appropriate locations, and told my users to have a nice day, turning what might have been a several-days-long recovery process into a several-hours-long "we'll test your ability to print when you get back from lunch," hand-wavy, user-happy-making experience.
You may question my abilities to administer systems based on that last paragraph, but in my defense the user base I am describing is several hundred independent offices, with users whose prowess ranges from "sure, I can ftp that to you when I get back to the office" to "what power button?" I wish I were kidding. To make things more interesting, the offices range across 4 states, and deal directly with municipal infrastructure, typically not in "major" cities. Try explaining "minimum downtime" to someone who isn't aware that a computer is required to access "that innernut thang".
Regardless of whether it's a server in a rack in some datacenter somewhere, or a re-purposed PC in your living room, the concepts are sound. Also, if you are already familiar with securely accessing the system remotely, you won't have to learn a whole new set of commands and procedures if/when you decide to go with a hosting service and/or move to the cloud.
(I'm guessing you're going to be physically sat at the computer while you're working on it, or working over VNC or some such, and that therefore you'll want a GUI - in this case use the usual Ubuntu desktop installer (rather than the -server edition which contains no GUI stuff).
A potentially critical difference you may not have considered is kernel optimization. The server version is optimized for "background processes", whereas the desktop is optimized for the "user experience"; these are two entirely separate concerns, and optimizing one will severely impact the other's performance.
A better solution might be to install the server, then "sudo apt-get install ubuntu-desktop", which will give you the desktop's GUI stuff without having the performance impact of the desktop's user-experience kernel optimizations.
The solution I have chosen is to use my preferred desktop OS on my preferred desktop machine, and interact with the server remotely via ssh and/or administration tools such as webmin or virtualmin. My servers have power and network attachment, and that's it. No keyboard, mouse, monitor, speakers, or anything else extraneous to their assigned tasks.
This has the added benefit of not requiring any procedural changes if/when you decide to move your server to the cloud, or otherwise outsource/co-locate it.
Also, if you are uneasy with a CLI, you need to ask yourself whether you can actually do this. Unease with CLI is typically due to lack of understanding how things actually work. In that case you should stay in the MS world, where people are (mostly) protected from making severe mistakes but are seriously limited in what they can actually do.
If you are trying to break out of the MS-world corset, then do not fear the CLI, learn to use it instead. It is the only way to be free of those restrictions, as GUIs are not and cannot be powerful interfaces due to fundamental limitations. Remember, the CLI gives you the power to command (and shoot yourself in the foot), while the GUI just interfaces you, allowing you to do just what the GUI designer chose to allow you to do. Both terms are surprisingly self-explanatory.
I wish I weren't so deeply embroiled in this discussion, so I could toss a mod point your way. This was an excellent response, directly addressing the poster's unstated needs. Kudos to you, sir.
An interesting (although completely wrong) viewpoint.
Nothing against you personally, but I find that people who spew party lines (such as "Ubuntu: an African word meaning (insert implication that user is retarded)") tend to have little or no experience with the distro itself, and are merely parroting information they picked up in some snarky forum in an attempt to appear to be somehow better than "the rest of the rabble".
That said, the Ubuntu forums are huge, bloated with reported issues because many desktop users (read: Windows-only users who wish to learn about their options) have been pointed at Ubuntu as an easy learning curve. On the other hand, how many of those problems go unsolved? The user base is large, but this means that solutions crop up almost as quickly as issues are brought to their attention.
A large amount of users engenders new potential issues that RTFM-shouting elitist assholes have no idea how to cope with, because only 6 or 7 people have managed to actually find and use their specific and precise (or home-rolled) distro of choice, so they have no idea how to support a larger user base, and are threatened if someone suggests there might be any issues whatsoever with their "perfect" platform.
In other news, when complaining that a specific solution is inappropriate, it is customary to offer a more appropriate solution; typically with supporting information as to why your preferred solution is better suited to a particular purpose.
Oh, and to bring you up-to-date: the distro flamewars died out several years ago. If you're going to troll, at least try to stay current and relevant.
Sorry about the wall of text... I'm new to slashdot posting and it seemed to have killed my formatting.
That's one of the reasons I choose to use the "plain text" option, and add in my own HTML if I feel the need. Also, you have now discovered the reason that the preview button exists.
Non-standard ports and/or fail2ban are good at reducing the chances of an attacker gaining access, but if you also use port knocking, you'll minimize the chances of an attacker even finding your "vulnerable" service. Multiple layers of security are better than single layers, hands down.
In todays environment, you need to understand a great deal about many OS level things, most of which revolve around security. Firewalls, mail configurations, etc. Not setting these things up correctly can have bad consequences. For example, your server IP address is blacklisted by RBLs because you left your SMTP port open and spammers started using it.
I'll respond to this with some EZ-Mode FixIt, and a small amount of snark directed at your pompous attitude.
Use a whitelist-based firewall (that is, only allow known legitimate connections (this can be per service/port rather than by remote IP, in case that wasn't obvious)). A firewall generation script can build you a good starting point, if you would prefer for it to be (mostly) done for you.
Use port knocking (look up knockd) for any "dangerous" services, such as SSH, as a further step in keeping the bad guys out of your box. No, this is not a good idea as the *only* security practice, but I have found that my SSH attacks have gone away completely from my logs since I implemented port knocking, whereas even port shifting and removing password-based authentication barely even slowed down the zombies blindly attacking my sshd (and fail2ban is useless for a distributed brute-force attack, such as the botnets that seem to do exactly that in their idle time).
Know that any decent (read: popular in the slightest) distribution of Linux will be fairly secure out-of-the-box, in that potentially dangerous services are typically not installed by default, and there are warnings about security all over the place when you attempt to install them. This means that your SMTP port scenario is an utter fabrication, because there's no mail server installed by default, and/or there's no configuration for it, so it's not running (or at least not accessible anywhere except 127.x.x.x). You are far more likely to be blacklisted because you're running it in your living room on your home's cable/dsl connection (via "public" IP range blacklisting) than because you accidentally left a port open on a default install.
No handy Linux app that I can see yet that replicates the user-friendly behavior of MacOS SuperDuper (sigh).
dd will probably do what you want. It's not exactly user-friendly by default, but there are tons of front-ends for it that make it more so, and I don't know of any *nix distributions that don't include it by default.
Technically speaking, dd "converts and copies a file"... but in *nix, everything is a file. That means disks, pipes, processes, and even, well, files.
While I see your point as to learning the ins-and-outs of your server and configuring it to be as efficient as is humanly possible, I find myself disagreeing with your post on a fundamental level.
While your method will probably produce a much leaner system, with slightly higher performance-per-clock-cycle, you are suggesting a system that requires a huge amount of "free time" to learn, and a willingness to dive head first into a tar pit of new knowledge. If you have a year or so to beat your head against a thousand brick walls while you climb a rather steep learning curve, then Gentoo is perfectly acceptable, and can (eventually) produce a leaner, cleaner server environment, tailored exactly to the task at hand.
If, on the other hand, you want your server to operate much like an appliance, then a simpler distribution (for example: something Debian-based that practically configures itself for you out of the box) can get you off the ground with enough knowledge to actually administer the thing, inside of a week... and throwing a multi-core CPU and a few gigs of ram at the hardware end of things is not only relatively inexpensive, but also pretty much nullifies Gentoo's "tuned performance" argument.
A cheap machine with enough power to be "good enough" (combined with a LAMP stack from any major distribution) will suffice, and doesn't require a masters in computer science to get running well enough to push a forum's backend.
* Note: I am assuming the user requesting the assistance is not familiar with Linux at all, thus increasing the potential learning curve steepness. Not everyone has been using *nix for 15 years, and not many people have the free time available to drop everything and immerse themselves in building a system from the ground up for several months of really intense self-education.
TL;DR: commandline-only interface on the server is fine, since you won't be administering the server locally in most cases. To implement: "sudo apt-get install webmin sshd knockd", then read the documentation and edit/create your configuration files.
Webmin is a useful tool for "avoiding the command line"; it gives you a browser-based interface to many common server systems and tools.
Between webmin and SAMBA, you can avoid the CLI for many common tasks, if that truly is your goal.
On the other hand, if you're serious about administering your server, you'll just bite the bullet and learn the handful of commands you'll need on the CLI to do the things you need to do, and read the man pages for ssh ("Secure SHell", a remotely accessible command-line interface using cryptographic security measures).
Implement port-knocking (Google "knockd"), use a non-standard SSH port, and implement certificate-based security to simplify your security concerns and keep the bots from being able to crack your sshd.
None of my servers have anything attached except power and network, unless/until there is a reason to interact directly with them - remote administration is the way to go.
Google forces review websites to provide their content for free to benefit Google’s own competing product – not consumers.
No, that would be the freely available content of the World Wide Web. Post it online, and it's available for anyone to see.
Google then gives its own product preferential treatment in Google search results.
... or maybe Google products get used by literally tens, if not hundreds, of millions of people all over the world, and so rank higher in the search results.
In 2010, Google began incorporating the content that it indexed from its competitors into Google Local without permission. Although Google had previously acknowledged that it needed a license to use Yelp’s content, it was now using it without permission to prop up its own, less effective product. In some instances, Google even presented this content to its users as if it were its own. [See Exhibit C]
In response to our objections, Google informed us that it would cease the practice only if we agreed to be removed from Google’s web search index, thereby preventing Yelp from appearing anywhere in Google web search results. This, of course, was a false choice. Google’s dominant position in the market prevents services like Yelp from exercising any sort of meaningful choice in the matter: it is a choice between allowing Google to co-opt one’s content and not competing at all.
So, let me get this straight... This guy is complaining about Google indexing his site, but then cries when Google offers to stop indexing his site, because it would reduce their web presence. In essence, he's mad that Google offered a fair exchange: either Google indexes the site and presents the information to its users, or Google doesn't index the site and the site has to flail about on its own to get more eyes for its content.
You can't eat your cookies while complaining that the chef who made them has his own cookies, too. Or rather, you can, but it makes you look like a spoiled brat who shouldn't have cookies in the first place.
We again asked that Google cease its practice of co-opting content from Yelp for its own benefit. Google responded by removing Yelp links from portions of Google’s web search product, providing a new twist on the same old false choice: if we chose not to help power Google Local, we could not appear in the “merged” portions of Google’s web search results.
Translation: We told them to stay off our site, and they stopped linking to our site. Gee, why would they do that?
Your argument is specious, damn near an outright fabrication, and your linked evidence is simply another company crying that Google made a better product using publicly-available resources. If you are trying to say the Internet is not a publicly-available resource, then I'm going to laugh at you.
What's happening with google is that they arguably have a monopoly on both the Online Advertising and Web Search markets.
No, they don't. If you have facts to back this statement up, then present them. Otherwise, I think we can safely discount this statement as fallacious.
Microsoft needs to advertise Bing in order to successfully challenge google on Web Search, but google is using its Online Advertising monopoly to prevent that -- which, to me, sounds like leveraging a monopoly (though this time around the leveraging is defensive, with Google trying to keep its search monopoly, rather than Microsoft's offensive leveraging aimed at conquering a position in the browser market)
Microsoft is Bing and other products. They have radio, TV, and print ads all over the place... and they have ads on Google, despite this supposed price hike. Google's ad system charges advertisers based on a bidding system, where advertisers spend more money for keywords that are requested more often. The price increase is a natural result of this "auctioning".
I think what he objects to is Google "cooking" the results to rank its own subsidiaries higher than competitor. It's actually quite similar to the MS antitrust case, where they have been accused of using their market position to corner the browser market.
Have you considered that Google's products rank so high on their (or any) search engine results because they have good products that hundreds of millions of people use?
This isn't antitrust. If you are using Google's services, then you have a choice immediately and obviously accessible; direct your browser to a different website.
We are not Google's customers (as lots of others have pointed out). Our attention is their product, and their customers are companies which wish to sell to Google's users.
I never said we were Google's "customers"... I stated that people use their products. In order to be clear, the products I was referring to were their services, such as search engine, office document suite, maps/navigation, operating systems, etc.
The issue at stake here is that a senator has proclaimed that Google is cooking their search results and Microsoft has cried foul over pricing, so now Google is getting in trouble for making good products. A similar complaint might be Jimmy Dean crying about Eggo not allowing them to put coupons in their waffle boxes.
As for the senator's complaints, if you watch the video of the interrogation^W interview, Schmidt repeated several times that showing Google coming up third in every result meant nothing, because they were "comparing apples to oranges" - comparing Google's search algorithm to other companies' price comparison engines.
Microsoft is not the only one affected by the advertising price increases, and the other companies who have sued have been quietly ushered out of the courtroom and told to quit whining. I see nothing improper about raising the price of a product, especially one which has pricing based on bidding on keywords. Microsoft is upset that the prices have gone up, but the increase is a result of natural market forces.
In short, this whole thing is a "media scandal" designed to capture everyone's attention and make Google out to be the bad guy for managing to make a profit with services rendered to consumers for free. I guess making a better product and selling it at a lower price than the competition is illegal, now.
This isn't antitrust. If you are using Google's services, then you have a choice immediately and obviously accessible; direct your browser to a different website. The Microsoft antitrust suits were more about them bundling IE with their OS, which forces the user to use it, even if it's only to download another browser. This activity, combined with the fact that it was incredibly difficult (some would say impossible) to purchase a PC at the time without a Windows(tm) license attached to it meant that they were leveraging their OS dominance to push their other software, which is how they got in trouble. If Google wants to link to Google services at the top of their search results, so be it. If Google wants to charge Microsoft one hundred million dollars for a single-line advertisement... hell, if Google wants to tell MSFT to go fly a kite, then so be it.
Last I checked, businesses were still able to define their own prices (in most cases), and to sell (or not sell) their products and services to whomever they want to.
Why should Google let MSFT advertise in the first place? This would be akin to a television station selling advertising space to a different television station.
Microsoft got slapped on the wrist for being a bully, and is now trying to be a tattletale and get the other kids in trouble.
-- "Sit them in the corner, mommy, they won't let me break their toys!"
You're saying Google shouldn't be allowed to market its own products on its own web pages, hosted by its own servers?
Or are you just objecting to them dropping a link at the top of the page to the Google version of whatever product or service you happen to have searched for?
If you have an issue with using Google products, then... well... don't.
Problem solved, and I didn't even hit you with a surcharge. Have a nice day.
... And here I thought it was going to be something along the lines of "We reserve the right to refuse service to anyone, for any reason" ; maybe with a healthy dose of IDFLY (I Don't F'ing Like You) fee.
Personally, I would charge Microsoft even more than Google has... "sorry, you're our competition; If you want to advertise with us, you'll need to pay the 'We don't like you' fee, the 'Microsoft sucks' fee, and a 'huh, thought you were the big boys' fee... for administrative purposes."
This is not anti-MSFT ranting, merely an observation of how I thought things worked (and if they don't, then I've got a civil suit against the local convenience store for throwing me out because I wasn't wearing shoes).
Slashdot Mirror
You can't predict the future of share prices by using past data.
No, but you can make a fairly-accurate educated guess about the price of Company X's share price tomorrow, based on the news today. The main issue about attempting to predict the future based on history is to know enough about the effect you are attempting to predict to understand its causes, and therefore be able to accurately detect the symptoms in time to react to them appropriately.
Of course, if we're detecting revolutions, then a fascist state would simply jail or otherwise dispose of anyone who performed any action they deemed to be a precursor to fomenting rebellion, thus removing any visible precursors. Of course, doing that might be the final issue that pushes the unrest over the cusp, and starts the revolution...
It's a sticky subject on a slippery slope.
algorithms - almost as useful as common sense
Or, "Algorithms - applied common sense for an unattended environment"
If I recall correctly, one of the postulates of Foundation was that the general populace should not be aware of the existence of the prediction mechanism, or it would fail in its purpose. Also, the groups of people whose actions were being predicted was so large that an individual person's actions should have little bearing on the results.
This concept has been summed up quite well by Agent K in Men In Black:
"A person is smart. People are stupid, panicky animals... and you know it."
Foundation anyone?
How about Paycheck?.
Philip K Dick is awesome, and his short stories have been made into many more movies than people suspect.
Asimov wasn't alone. Several authors have put forth the concept (my favorite is actually "Cleology", introduced in "In the Country of the Blind" by Michael Flynn, complete with some graphs and charts of historical cyclical data that projected (at the time of the book's writing) future events that have turned out to be surprisingly accurate (at the time of this writing)).
The entire concept of studying history can be summed up in the phrase "Those who do not recall history are doomed to repeat it" - in other words, we need to know what mistakes we have made in order to avoid making them again.
Tracking cookies track. This is not news, this is anticipated and expected behavior. This has been the status quo for over a decade.
Cookies have a security feature in that they are accessible only to the websites that placed them, but advertising sites have been using tracking cookies for as long as cookies have existed, and getting around that security by placing a "bug" on third-party sites. They used to (and probably still do) implement this as a 1x1 "spacer" image the same color as the background, or simply by having an ad on the page you are viewing. When your browser requests the image/flash/javascript/whatever, the site it comes from is suddenly allowed to access their cookie.
The solution has also not changed; either don't allow cookies, or delete them constantly. Anti-scripting addons are also helpful, as are black (or whitelists) of websites to disallow (or allow) access to your system. Modifying hosts files has been a semi-successful method, as well, in that requests sent to specific named addresses can be redirected to localhost (and therefore "blocked").
I personally use NoScript and AdBlockPlus for precisely this reason (and to speed up my page loads), and I can't fathom why this information could be conceived to be news to any user with any amount of technical knowledge and a modicum of interest in their own privacy.
The paragraph in my above post that starts "As for firing up dd on the raw device" should have included "on a currently active system" for the rest of that paragraph to be valid and make sense.
I was pretty sure that dd was going to be my best friend, but we could wish for a better best friend. As near as I can tell, I ought to boot single-user off a CD and do it from there. Is that overconservative? No-way-no-how can I just far up dd on the raw device underneath a bunch of partitions, can I?
Look for dd-based partition-manipulation tools; possible search terms might be "Linux Ghost" or "disk image dd". There are a plethora of solutions out there, and you can probably find one that suits you with only a small amount of experimentation. Many are available as a bootable CD, and come with a menu system or other helpful tools. Several allow you to backup to a massive variety of targets, anything from simply backing up a partition to a gzip'd archive on a separate partition to streaming a bit-wise copy of the entire disk over the internet to a remote system you have access to (and creating an exact duplicate of the partition/drive on that target).
A LiveCD of your favorite distro or "rescue disk" or "recovery tools" can also be extremely useful; Most either already contain some sort of disk/partition tool or allow it to be installed on the RAMdrive the system boots into, assuming internet access. For example, gparted is a graphical tool that will allow you to manipulate partitions, resizing, moving, cloning, etc... and it's available on the Ubuntu LiveCD by default (last I checked).
A persistence-capable "external" (USB or otherwise) boot device can allow you to keep configuration information or notes (such as which drives/partitions you have backed up, when and to where) between boots, as well as allowing a portable system that contains your preferred set of tools. Taken a step further, you can create your own LiveCD once you have a preferred setup (assuming you're not scared to learn even more).
As for firing up dd on the raw device underneath a bunch of partitions... well, your swap will be corrupted, for obvious reasons, and any disk writes during the read process will harf your data, too, so it's not necessarily the best plan... but it's theoretically possible. At this point, I would examine your motives for imaging the disks in the first place, as a direct image may actually take more time than simply backing up your configuration and data files, and keeping a list of where each file belongs and what packages need to be installed if the server should be "hit by a bus".
Disk imaging can be a simple solution, but a reinstall can be an excellent opportunity for "cleaning out the cruft", and may actually require less time or effort than imaging a full disk, depending on the size of the data involved and where the server is located (ie, your ability to gain physical access). I've lost count of how many times I have blown a new OS onto a machine, dropped the appropriate apps/data from backup into the appropriate locations, and told my users to have a nice day, turning what might have been a several-days-long recovery process into a several-hours-long "we'll test your ability to print when you get back from lunch," hand-wavy, user-happy-making experience.
You may question my abilities to administer systems based on that last paragraph, but in my defense the user base I am describing is several hundred independent offices, with users whose prowess ranges from "sure, I can ftp that to you when I get back to the office" to "what power button?"
I wish I were kidding. To make things more interesting, the offices range across 4 states, and deal directly with municipal infrastructure, typically not in "major" cities. Try explaining "minimum downtime" to someone who isn't aware that a computer is required to access "that innernut thang".
Regardless of whether it's a server in a rack in some datacenter somewhere, or a re-purposed PC in your living room, the concepts are sound. Also, if you are already familiar with securely accessing the system remotely, you won't have to learn a whole new set of commands and procedures if/when you decide to go with a hosting service and/or move to the cloud.
(I'm guessing you're going to be physically sat at the computer while you're working on it, or working over VNC or some such, and that therefore you'll want a GUI - in this case use the usual Ubuntu desktop installer (rather than the -server edition which contains no GUI stuff).
A potentially critical difference you may not have considered is kernel optimization. The server version is optimized for "background processes", whereas the desktop is optimized for the "user experience"; these are two entirely separate concerns, and optimizing one will severely impact the other's performance.
A better solution might be to install the server, then "sudo apt-get install ubuntu-desktop", which will give you the desktop's GUI stuff without having the performance impact of the desktop's user-experience kernel optimizations.
The solution I have chosen is to use my preferred desktop OS on my preferred desktop machine, and interact with the server remotely via ssh and/or administration tools such as webmin or virtualmin. My servers have power and network attachment, and that's it. No keyboard, mouse, monitor, speakers, or anything else extraneous to their assigned tasks.
This has the added benefit of not requiring any procedural changes if/when you decide to move your server to the cloud, or otherwise outsource/co-locate it.
Also, if you are uneasy with a CLI, you need to ask yourself whether you can actually do this. Unease with CLI is typically due to lack of understanding how things actually work. In that case you should stay in the MS world, where people are (mostly) protected from making severe mistakes but are seriously limited in what they can actually do.
If you are trying to break out of the MS-world corset, then do not fear the CLI, learn to use it instead. It is the only way to be free of those restrictions, as GUIs are not and cannot be powerful interfaces due to fundamental limitations. Remember, the CLI gives you the power to command (and shoot yourself in the foot), while the GUI just interfaces you, allowing you to do just what the GUI designer chose to allow you to do. Both terms are surprisingly self-explanatory.
I wish I weren't so deeply embroiled in this discussion, so I could toss a mod point your way. This was an excellent response, directly addressing the poster's unstated needs. Kudos to you, sir.
An interesting (although completely wrong) viewpoint.
Nothing against you personally, but I find that people who spew party lines (such as "Ubuntu: an African word meaning (insert implication that user is retarded)") tend to have little or no experience with the distro itself, and are merely parroting information they picked up in some snarky forum in an attempt to appear to be somehow better than "the rest of the rabble".
That said, the Ubuntu forums are huge, bloated with reported issues because many desktop users (read: Windows-only users who wish to learn about their options) have been pointed at Ubuntu as an easy learning curve. On the other hand, how many of those problems go unsolved? The user base is large, but this means that solutions crop up almost as quickly as issues are brought to their attention.
A large amount of users engenders new potential issues that RTFM-shouting elitist assholes have no idea how to cope with, because only 6 or 7 people have managed to actually find and use their specific and precise (or home-rolled) distro of choice, so they have no idea how to support a larger user base, and are threatened if someone suggests there might be any issues whatsoever with their "perfect" platform.
In other news, when complaining that a specific solution is inappropriate, it is customary to offer a more appropriate solution; typically with supporting information as to why your preferred solution is better suited to a particular purpose.
Oh, and to bring you up-to-date: the distro flamewars died out several years ago. If you're going to troll, at least try to stay current and relevant.
Sorry about the wall of text... I'm new to slashdot posting and it seemed to have killed my formatting.
That's one of the reasons I choose to use the "plain text" option, and add in my own HTML if I feel the need.
Also, you have now discovered the reason that the preview button exists.
Non-standard ports and/or fail2ban are good at reducing the chances of an attacker gaining access, but if you also use port knocking, you'll minimize the chances of an attacker even finding your "vulnerable" service. Multiple layers of security are better than single layers, hands down.
In todays environment, you need to understand a great deal about many OS level things, most of which revolve around security. Firewalls, mail configurations, etc. Not setting these things up correctly can have bad consequences. For example, your server IP address is blacklisted by RBLs because you left your SMTP port open and spammers started using it.
I'll respond to this with some EZ-Mode FixIt, and a small amount of snark directed at your pompous attitude.
Use a whitelist-based firewall (that is, only allow known legitimate connections (this can be per service/port rather than by remote IP, in case that wasn't obvious)). A firewall generation script can build you a good starting point, if you would prefer for it to be (mostly) done for you.
Use port knocking (look up knockd) for any "dangerous" services, such as SSH, as a further step in keeping the bad guys out of your box. No, this is not a good idea as the *only* security practice, but I have found that my SSH attacks have gone away completely from my logs since I implemented port knocking, whereas even port shifting and removing password-based authentication barely even slowed down the zombies blindly attacking my sshd (and fail2ban is useless for a distributed brute-force attack, such as the botnets that seem to do exactly that in their idle time).
Know that any decent (read: popular in the slightest) distribution of Linux will be fairly secure out-of-the-box, in that potentially dangerous services are typically not installed by default, and there are warnings about security all over the place when you attempt to install them. This means that your SMTP port scenario is an utter fabrication, because there's no mail server installed by default, and/or there's no configuration for it, so it's not running (or at least not accessible anywhere except 127.x.x.x). You are far more likely to be blacklisted because you're running it in your living room on your home's cable/dsl connection (via "public" IP range blacklisting) than because you accidentally left a port open on a default install.
No handy Linux app that I can see yet that replicates the user-friendly behavior of MacOS SuperDuper (sigh).
dd will probably do what you want. It's not exactly user-friendly by default, but there are tons of front-ends for it that make it more so, and I don't know of any *nix distributions that don't include it by default.
Technically speaking, dd "converts and copies a file"... but in *nix, everything is a file. That means disks, pipes, processes, and even, well, files.
While I see your point as to learning the ins-and-outs of your server and configuring it to be as efficient as is humanly possible, I find myself disagreeing with your post on a fundamental level.
While your method will probably produce a much leaner system, with slightly higher performance-per-clock-cycle, you are suggesting a system that requires a huge amount of "free time" to learn, and a willingness to dive head first into a tar pit of new knowledge. If you have a year or so to beat your head against a thousand brick walls while you climb a rather steep learning curve, then Gentoo is perfectly acceptable, and can (eventually) produce a leaner, cleaner server environment, tailored exactly to the task at hand.
If, on the other hand, you want your server to operate much like an appliance, then a simpler distribution (for example: something Debian-based that practically configures itself for you out of the box) can get you off the ground with enough knowledge to actually administer the thing, inside of a week... and throwing a multi-core CPU and a few gigs of ram at the hardware end of things is not only relatively inexpensive, but also pretty much nullifies Gentoo's "tuned performance" argument.
A cheap machine with enough power to be "good enough" (combined with a LAMP stack from any major distribution) will suffice, and doesn't require a masters in computer science to get running well enough to push a forum's backend.
* Note: I am assuming the user requesting the assistance is not familiar with Linux at all, thus increasing the potential learning curve steepness. Not everyone has been using *nix for 15 years, and not many people have the free time available to drop everything and immerse themselves in building a system from the ground up for several months of really intense self-education.
TL;DR: commandline-only interface on the server is fine, since you won't be administering the server locally in most cases.
To implement: "sudo apt-get install webmin sshd knockd", then read the documentation and edit/create your configuration files.
Webmin is a useful tool for "avoiding the command line"; it gives you a browser-based interface to many common server systems and tools.
Between webmin and SAMBA, you can avoid the CLI for many common tasks, if that truly is your goal.
On the other hand, if you're serious about administering your server, you'll just bite the bullet and learn the handful of commands you'll need on the CLI to do the things you need to do, and read the man pages for ssh ("Secure SHell", a remotely accessible command-line interface using cryptographic security measures).
Implement port-knocking (Google "knockd"), use a non-standard SSH port, and implement certificate-based security to simplify your security concerns and keep the bots from being able to crack your sshd.
None of my servers have anything attached except power and network, unless/until there is a reason to interact directly with them - remote administration is the way to go.
From the text you linked to:
Google forces review websites to provide their content for free to benefit Google’s own competing product – not consumers.
No, that would be the freely available content of the World Wide Web. Post it online, and it's available for anyone to see.
Google then gives its own product preferential treatment in Google search results.
... or maybe Google products get used by literally tens, if not hundreds, of millions of people all over the world, and so rank higher in the search results.
In 2010, Google began incorporating the content that it indexed from its competitors into Google Local without permission. Although Google had previously acknowledged that it needed a license to use Yelp’s content, it was now using it without permission to prop up its own, less effective product. In some instances, Google even presented this content to its users as if it were its own. [See Exhibit C]
In response to our objections, Google informed us that it would cease the practice only if we agreed to be removed from Google’s web search index, thereby preventing Yelp from appearing anywhere in Google web search results. This, of course, was a false choice. Google’s dominant position in the market prevents services like Yelp from exercising any sort of meaningful choice in the matter: it is a choice between allowing Google to co-opt one’s content and not competing
at all.
So, let me get this straight... This guy is complaining about Google indexing his site, but then cries when Google offers to stop indexing his site, because it would reduce their web presence. In essence, he's mad that Google offered a fair exchange: either Google indexes the site and presents the information to its users, or Google doesn't index the site and the site has to flail about on its own to get more eyes for its content.
You can't eat your cookies while complaining that the chef who made them has his own cookies, too. Or rather, you can, but it makes you look like a spoiled brat who shouldn't have cookies in the first place.
We again asked that Google cease its practice of co-opting content from Yelp for its own benefit. Google responded by removing Yelp links from portions of Google’s web search product, providing a new twist on the same old false choice: if we chose not to help power Google Local, we could not appear in the “merged” portions of Google’s web search results.
Translation: We told them to stay off our site, and they stopped linking to our site. Gee, why would they do that?
Your argument is specious, damn near an outright fabrication, and your linked evidence is simply another company crying that Google made a better product using publicly-available resources. If you are trying to say the Internet is not a publicly-available resource, then I'm going to laugh at you.
What's happening with google is that they arguably have a monopoly on both the Online Advertising and Web Search markets.
No, they don't. If you have facts to back this statement up, then present them. Otherwise, I think we can safely discount this statement as fallacious.
Microsoft needs to advertise Bing in order to successfully challenge google on Web Search, but google is using its Online Advertising monopoly to prevent that -- which, to me, sounds like leveraging a monopoly (though this time around the leveraging is defensive, with Google trying to keep its search monopoly, rather than Microsoft's offensive leveraging aimed at conquering a position in the browser market)
Microsoft is Bing and other products. They have radio, TV, and print ads all over the place... and they have ads on Google, despite this supposed price hike. Google's ad system charges advertisers based on a bidding system, where advertisers spend more money for keywords that are requested more often. The price increase is a natural result of this "auctioning".
I think what he objects to is Google "cooking" the results to rank its own subsidiaries higher than competitor. It's actually quite similar to the MS antitrust case, where they have been accused of using their market position to corner the browser market.
Have you considered that Google's products rank so high on their (or any) search engine results because they have good products that hundreds of millions of people use?
This isn't antitrust. If you are using Google's services, then you have a choice immediately and obviously accessible; direct your browser to a different website.
We are not Google's customers (as lots of others have pointed out). Our attention is their product, and their customers are companies which wish to sell to Google's users.
I never said we were Google's "customers"... I stated that people use their products. In order to be clear, the products I was referring to were their services, such as search engine, office document suite, maps/navigation, operating systems, etc.
The issue at stake here is that a senator has proclaimed that Google is cooking their search results and Microsoft has cried foul over pricing, so now Google is getting in trouble for making good products. A similar complaint might be Jimmy Dean crying about Eggo not allowing them to put coupons in their waffle boxes.
As for the senator's complaints, if you watch the video of the interrogation^W interview, Schmidt repeated several times that showing Google coming up third in every result meant nothing, because they were "comparing apples to oranges" - comparing Google's search algorithm to other companies' price comparison engines.
Microsoft is not the only one affected by the advertising price increases, and the other companies who have sued have been quietly ushered out of the courtroom and told to quit whining. I see nothing improper about raising the price of a product, especially one which has pricing based on bidding on keywords. Microsoft is upset that the prices have gone up, but the increase is a result of natural market forces.
In short, this whole thing is a "media scandal" designed to capture everyone's attention and make Google out to be the bad guy for managing to make a profit with services rendered to consumers for free. I guess making a better product and selling it at a lower price than the competition is illegal, now.
This isn't antitrust. If you are using Google's services, then you have a choice immediately and obviously accessible; direct your browser to a different website. The Microsoft antitrust suits were more about them bundling IE with their OS, which forces the user to use it, even if it's only to download another browser. This activity, combined with the fact that it was incredibly difficult (some would say impossible) to purchase a PC at the time without a Windows(tm) license attached to it meant that they were leveraging their OS dominance to push their other software, which is how they got in trouble. If Google wants to link to Google services at the top of their search results, so be it. If Google wants to charge Microsoft one hundred million dollars for a single-line advertisement... hell, if Google wants to tell MSFT to go fly a kite, then so be it.
Last I checked, businesses were still able to define their own prices (in most cases), and to sell (or not sell) their products and services to whomever they want to.
Why should Google let MSFT advertise in the first place? This would be akin to a television station selling advertising space to a different television station.
Microsoft got slapped on the wrist for being a bully, and is now trying to be a tattletale and get the other kids in trouble.
--
"Sit them in the corner, mommy, they won't let me break their toys!"
So, let me see if i understand you properly...
You're saying Google shouldn't be allowed to market its own products on its own web pages, hosted by its own servers?
Or are you just objecting to them dropping a link at the top of the page to the Google version of whatever product or service you happen to have searched for?
If you have an issue with using Google products, then... well... don't.
Problem solved, and I didn't even hit you with a surcharge. Have a nice day.
... And here I thought it was going to be something along the lines of "We reserve the right to refuse service to anyone, for any reason" ; maybe with a healthy dose of IDFLY (I Don't F'ing Like You) fee.
Personally, I would charge Microsoft even more than Google has... "sorry, you're our competition; If you want to advertise with us, you'll need to pay the 'We don't like you' fee, the 'Microsoft sucks' fee, and a 'huh, thought you were the big boys' fee... for administrative purposes."
This is not anti-MSFT ranting, merely an observation of how I thought things worked (and if they don't, then I've got a civil suit against the local convenience store for throwing me out because I wasn't wearing shoes).