Slashdot Mirror
Microsoft Claims Firms 'Hitting a Wall' With Linux
maxifez writes writes to tell us that Microsoft has released yet another independent study downplaying the viability of Linux at the enterprise level. The study claims that Windows is "more consistent, predictable, and easier to manage than Linux." From the article: "The study, commissioned by the software giant from Security Innovation, a provider of application security services, claimed that Linux administrators took 68 per cent longer to implement new business requirements than their Windows counterparts." Vnunet.com has also provided a PDF of the original report.
Also I am pleased to notice that the "independant company" that spewed out this "commissioned report" (see the microsoft page) lists Microsoft on their partners page, and from what I can tell no one who has even heard of Linux (with the exception of HP). Since it looks like their business depends on selling enhanced security products, I can see why they wouldn't be too keen on having people show an interest in Linux.
Security Innovation designed this study to be repeatable, and we believe that the results are consistent with what customers are experiencing in the real world.
And if I wander over and look at my main Linux file server, I see an uptime of 125 days (it had over a year uptime before I physically had to move the server to a different location). During that time the server's files were available 24/7 with absolutely zero problems. Needless to say we have had way fewer problems with the new Linux server; the old Microsoft server crashed or had to be rebooted on a regular basis; the people before me actually had a planned "weekly reboot every Friday evening". When it came time to replace the Microsoft server, Microsoft didn't fare so well, especially when it came time to pay big piles of $$$ to upgrade it: basically we could have bought two linux file servers for the cost of the Microsoft software upgrade costs alone. Hows that for a real-world example?
I Am My Own Worst Enemy
The study claimed that Linux administrators took 68 per cent longer to implement new business requirements than their Windows counterparts.
yeah, maybe true. But how about maintaining them later, for years, with zero downtime?
#
#\ @ ? Colonize Mars
#
And I suppose the pro-Linux studies are more valid?
According to Netcraft:
My Websites Hosted on Linux:
Last Reboot: 468 days
Last Reboot: 331 days
Last Reboot: 664 days
Other of My Websites Hosted on Windows:
Last Reboot: 3 days
Last Reboot: 9 days
Last Reboot: 11 days
Customers wanting to switch from Windows to Linux: 3
99.999% Uptime and 50% happy customers: Priceless
CP
um, "68% of implementation time" has nothing to do with subsequent uptime. I would agree that implementing a new service on windows is considerably easier if for no other reason than there probably is a single product you can just install. On linux a single service may require cobbling together many different components. Go try and install something as simple as a calendar system in linux....you have to install ldap for authentication, a webserver for web access, the calendar software itself, the postgres database for data, etc. On windows you often just find a single app which does it all. For small to medium sized businesses, cobbling together doesn't make sense.
"As they attempt to increase business capabilities over time, customers are telling us that they are hitting a wall with Linux" said Martin Taylor, general manager of platform strategy at Microsoft.
I am sorry but this is the most sorry ass downhill marketing I have ever seen. It is like Pepsi hiring a guy to wear a Coca-Cola Tshirt and purposely choking. This guy changed title 3 times in a year. Just a couple months ago he was Microsoft's very own Linux strategist.
4 red hat 7.3 DNS servers. Have never required a reboot since installation.
1 red hat 6 machine that lasted 6 years without an OS related reboot (the hardware started to give and the box had to be decommissioned)
1 database/web server running SLES 8 has gone over a year and a half without a reboot.
1 webserver running debian stable no reboot since installation
1 proxy server running SLES 9 w/ squid that was set up in under a 1/2 hour under emergency conditions (old proxy hardware died unexpectedly) running 20 days under extremely heavy load until new server came in.
1 database server running SLES 8. A year since last reboot.
And those are all the ancient boxes. We've got many more linux boxen that are too new to have aquired a long uptime.
From the article:
experiencing significant reliability issues resulting in higher total cost of ownership
*shrug* I've had none of these issues they speak of. All of our installs are quick, stable and long lasting. In fact, I've never had a production upgrade break anything, and never had an install take longer than a couple of hours in even the most complex of setups.
This whole "get the facts" campaign is just silly. I don't know why they keep on with it. I've been working with Linux for years and never run into any of the problems they have "documented".
Hey Balmer, want an anecdotal story of Windows breaking? Our mapping department had a Windows 2000 installation with their mapping software. One day it just breaks. 5 people standing around the box scratching our heads. No one had any clue why. Random reboots, blue screens, the whole works. We reinstall many times. Nothing. Do all the upgrades, patches and fixes. Nothing. Sounds like hardware, right? Nope. Upgraded to 2003 and worked fine since.
The fact that the box could have run 2 years without major issue then break out of nowhere with 5 very smart people trying to solve the issue and can't makes me wonder.
Get the real facts.
If an officer ever threatens to taze you, say you have a pacemaker.
We're finding it very difficult use java with Windows in a production environment. We can't get a heap larger than 1GB without jumping through hoops. Even then, getting 1.5GB is about as good as it gets. For an application server environment with multiple integration points, this just isn't enough. On Linux or Solaris I can get just under 4GB just by flipping a switch. To get the 1.5GB or closer to 2GB on windows, you've got to disable all kinds of services and just hope you can get a contiguous region large enough.
... is a relative term I could compile a report like this demonstrating that Linux admins take 68% longer to perform a set of cherry picked tasks you can do alot faster on a Windows machine that has a nice easy to use GUI management tool specially designed to do those same tasks. I could also demonstrate to you that Windows admins take 68% longer to perform certain cherry picked tasks because those same Windows GUI management tools (Windows command line tools tend to suck ass) simply don't enable you to perform those tasks as efficiently as you can by doing them with shell/perl scripts on the Linux command line. Comparing Windows to Linux/Unix is to some extent akin to comparing cats and dogs. The design philosophies of Linux/Unix are fundamentally different from those of Windows. The former are meant to be more flexible and targeted at better educated operators while Windows seems to be geared firstly towards corporations who want something that a relatively low skilled person, preferably without a high level of education (and thus a lower salary), can easily administrate and secondly it is geared towards the mostly clueless average consumer. Then there is OS.X which does an admirable job of being just as easy, if not easyer to use than Windows, (while still being more secure) but it still has all the power of Linux/Unix making it a nice compromise.
Only to idiots, are orders laws.
-- Henning von Tresckow
Yea, I've got a win2k3 box hooked up as a Raster Image Processor...Basically it processes a binary image, makes it suitable for highrez film printing, does color separation, stuff like that. It takes a moderate amount of traffic, and needs to be rebooted at least once a week. Sometimes it goes into crash cycles on Fridays when the traffic is highest, and needs to be rebooted hourly.
I've got three antiquated Solaris boxes running older versions of the same software, and taking MORE traffic, that need to be rebooted about once a month.
I've gotta say, the software must be brutal because I've never had a Solaris machine have that many problems. Even so, Windows shows it's true colors as usual. On new hardware, with new software and all the patches, it's much less reliable than much older machines running a better os.
ad logicam Claiming a proposition is false because it was presented as the conclusion of a fallacious argument.
Let's say that Microsoft funded many fair and impartial studies and, let's say, 5% came to a pro-Microsoft conclusion. Which findings get published and advertised by Microsoft? Does that make the published findings unfair or impartial?
I'm not picking sides, but everyone is picking on the independent institute. Maybe they are independent.
Bah!
Now simply answer this. When did MS provided Ghost with install CD?
And yes it is easier with kickstart than ghost.
You still have to change names, ip, etc... Kickstart option can be selective. Ghost not.
Signature Pro version 1.13.2-3 release 83.5 beta3try7 after-breakfast edition
Well, just a few things, Active Directory is integrated, vs. NDS or other directory services, the fact that file permissions are more fine grained in ntfs vs. unix permissions... the setup wizards for things like IIS configuration leave less time figuring out configuration files, and more time doing the work...
Now, on the flip side, I've seen unix gurus that can configure apache without blinking. And can configure a lot of the security and network settings with ease... network configuration is much more customizable in *nix. Each do have advantages, and disadvantages.. but the ease of use in windows servers is a definative advantage... red hat enterprise servers (as well as suse) are pretty close, and in some areas, getting better.
On broad desktop deployments, SMS is really nice to have... Where MS's tools really shine is on the developer tools, and deployment using windows based servers... Oracle compared to MS' SQL Server, or DB2 isn't a comparison on ease of use/administration. But hey, I like both.. I'm a pretty big fan of PC/FreeBSD and like Linux... I *use* windows though.. I do wish security concerns took more of a front seat though.
Michael J. Ryan - tracker1.info
Also, I never found any mention the 68% figure quoted in the article. In fact, there is no mention of the subject of upgrading in support of business needs at all; it deals only with deployment/use of a database server. Not to mentioned that the Windows system in the study is Server 2003, not Windows 2000, as is mentioned in the article.
In fact, now that I think about it, they MUST have linked to the wrong .pdf, because the study I read (linked above AND in the original article) have nothing to do with the statsitics quoted. (The link I followed is here.) Further, the .pdf was written in June 2005, but the article is from November...something is up...
Looking at the Microsoft site, they make no mention of those statistics either in regards to the linked report. Their summary page is here and links back to the study the article links to.
In fact, I did several google searches (and MS searches!) to try to find ANY study in the Microsoft whitepapers section that has any mention of upgrading from SUSE Linux 8 to 9, or a mention of "hitting the wall", or a mentioned on Linux taking 68% longer than Windows to upgrade to suit business needs. I found no results. In fact, there was only one result on the entire Microsoft site regarding SUSE Enterprise Linux 9, and it wasn't relevent. Oh, and Vnunet didn't provide a copy of the report as the submitter states...that link goes back to an (incorrect) page at Microsoft.
In short, can I read the study that this article is talking about? I'm curious to see what it says... =)
They had a year to study this and came up with some unusual metrics to say the least. For those who did not read the 44 pages of PDF, let me summarize some of my observations:
- They appear to be more comfortable with Windows than Linux. There is nothing wrong with that except they do not account for it in the time to complete tasks.
- They compared a Windows box running MS SS against two versions of Red Hat running MySQL and Oracle. That the did not use the same data bases on both OS slants the numbers from the start. Even if they wanted to avoid MySQL, they could have selected an Oracle installation.
- They counted vulnerabilities at the component level. So a shared library that had a vulnerability, but was used by both the installed OS and the database is counted twice. One used by the OS, the GUI, and the database, three times, etc. They state this is fair, but this would automatically penalize a Linux distribution because MS does not get counted twice in any case.
- The Red Hat installations were done manually and minimal installations. They then had problems, and make commentary on the difficulty of the upgrades. I would be very interested in the detail of what they did for the install. This appears to be a self-inflicted wound claiming to be otherwise.
- They make an big deal about what ports are open in the default installation. They comment that MS continues to allow MSUpdate, a good thing, but that Linux left the port for up2date open, a bad thing. Again, as a minimalist install they should have secured the ports, but that is dumb argument regardless. Admins who leave a machine wide open deserved to be fired. Because MS now ships theirs with everything closed is a side effect of the number of complaints about bad admins leaving the server in its out of the box state.
- Days to resolve a vulnerability are dangerous guides. First, a vulnerability has to be reported, then verified. We are dependent upon the vendor (MS, Oracle, etc) to correctly reflect these. However, almost anyone can and does report one for OSS - and that is a good thing.
In general, they speak of vulnerabilities and the ability to respond to business requests. I would like to see the requests they specifically refer to. While 68% sounds like a lot, is it the difference between 12 and 26 seconds? I just cannot see in my day to day activities it taking me more than half again as long to do anything and it is far less to image entire boxes. I wonder if this is a familiarity thing.
It is really time someone from RedHat or SUSE took a study like this and dissected it for a comparison 1:1 with MS. None of this it counts twice or differing databases garbage, a real compare. The top 20 tasks an admin will perform in a year. If we loose at least we know what to focus our energies upon. (What does not kill us, makes us stronger)
Supporting a mixed (Windows/Linux/Solaris) environment, I just do not see a 68% difference anywhere for an experienced admin.
Is it just me or are others sick of hearing about so called admins having really stable linux systems but thier windows systems always crash. There are PLENTY of windows server environments that don't crash, have AD configured right so that clients can't do stupid things, have automated features for antivirus/update/install roll outs. you just have to know what programs you need and how to use them, just like with linux. if your server is always crashing then you are a horrible admin or you installed bad 3rd party software. If your clients are always crashing then you didn't setup AD correctly or installed bad 3rd party software. I don't doubt that linux works well for many things. (though the thought of running servers with kde or gnome on is shocking) it love testing new linux distros too but you have to say that sometimes it's a pain to do things in linux that are easy to do in windows. support for hardware has gotten better but it's no where are good as windows. sure linux is better for somethings too but we all know this article isn't going to focus on that. let it rest at that instead of just showing us how badly you suck at being a windows admin.
My biggest problem with the article is the claim to "predictability and consistency". They probably haven't seen any of the past 20 AD implementations. I have seen AD, Windows 2000, and Windows 2003 show extreme unpredictability and terrible consistency. I have seen some crazy GPO get applied to users out of the blue, and I've seen some of the craziest errors ever. I think linux has the predictability and consistency, however, there is a little bit of upkeep required and a little more well trained tech staff, but hey, you get what you pay for. Deal with it.
YOU'RE WINNER !
Another lame blog
Interesting how these experts always bring up all that shit over and overjust before the end of the year ...
... soo tired of it seriously ... then 10 more articles appear: no windows sucks, no linux sucks ....
.. oh we are an application security firm and we want to sell you security software for every powerhouse that sits on every secretary's desk... it is really needed there ...
...
.....
does it have to do something about closing next years budget plans in e.g. US companies of a certain size ?
Windows is superior, Linux sucks
Linux took 68 percent longer to implement.... it was more secure, and did not have hours of downtime, and had instant patches, but hey bill send me the check and we don't write about that
Bored of it....
Probably I will be flamed to hell by various windows admins... and maybe they are right, I am more bored of the subject itself than the always biased "facts in them"
It is like 2 kids in the sandbox fighting over who's dad is stronger.....
The sad thing is that companies actually read these, and then make decisions,
One company I was in contact with was pretty open about it: we sell the solution which costs more and the one we can charge more for service costs..... in fact we try to sell Linux + Windows + some proprietary crap, in case they have an admin for some we have a fair chance of billing them for the other
Oooh, I better go back and work, my real point was: are these effecting "end of the year" decisions and does anyone think that is a systematical way to purchase a few more boxed windows cds or servers in the last minute?
Ghost is only "easy" if the machines are 100% identical. Ghost is a pain in the ass if there are subtle driver differences (chip sets, SCSI vs. SATA vs. IDE, etc). About 50% of the time I try to ghost configurations between 2 non-identical machines the system blue-screens and won't even boot to "Safe Mode"; "repair" doesn't even help. Give me kickstart over that madness any day.
Bullshit. Sun, IBM, and Novel are marketing Linux to the very high end of the spectrum. If you want pro-Linux studies, do a Google search.
Linux is a grassroot effort like the anti-tobacco movement -- both are backed by many millions of dollars.
OK, it's time for me to repost what's involved installing a private certificate server on Windows 2000 via its "intuitive" point-and-click GUI. (You forgive me if I just link to it, not wanting to repeat slashdot's lameness filter hell for this kind of post.) It compares the Linux way and the Windows way. These were the actual procedures used, that I carefully documented, for two different projects that accomplished exactly the same goal. Here it is. (Scroll past the lameness filter stuff at the beginning.)
"more consistent, predictable, and easier to manage than Linux."
I know I can manage thousands of servers with Active Directory. How do you do that in Linux? Is there some "out of the box" and standardized tool? I dont know Linux, so thats an actual question.
Windows has a ton of benefits over Linux, just from my 15 years in Corporate America (as a tech). It's easier to hire people who know how to manage Windows well. It's easier to have people support it. It's easier to get support if something goes wrong, and if you have good people, things rarely go wrong.
I have only seen one linux solution in the last serveral years. It was actually not too bad given it's situation: it was a replacement for a really big and really old Unix server, which really ran a database. Unfortunately, it's uptime wasnt all that impressive, and every time something went wrong an overpriced consultant had to be called in. And yes, it's anecdotal, but I hear this anecdotal situation repeated all the time.
Personally, I would have taken the hit and payed somebody to convert the database to something which would run on a Windows Server. Heck, we even had a Windows server managing the backups on that Linux server! Last I heard, that server was still breaking down occasionally; the worst thing was that particular database came as close to needing 24/7 as anything there.
It isnt going to gain me any brownie points around here, but from my experience, Linux is way overrated. I've had Windows NT4 servers with some damn great uptime, and they were running Exchange and being used close to 24/7.
You have to remember, Microsoft's customers use Windows, and it's a hell of a learning curve to go from a Windows GUI admin to a Linux one. I'm not surprised there 'hitting a brick wall' because there already standing at it's foot gazing to the stars.
I tried out Lindows the other day (just for fun) and it didn't setup dhcp on my wireless card, absolutely everything else worked perfectly! To a GUI Microsoft admin this would be like a brick wall, the GUI tool provided by Lindows did't let you setup dhcp from the boot cd, for a linux guy it's just a dhcpdc wlan0 away.
thank God the internet isn't a human right.
Just installed Windows XP Pro at work. I'm not addressing technical issues here, just theoretical ones. The End User License Agreement lists twenty seven ways in which Windows leaks data by default just to the Microsoft servers
This doesn't mention the others... By visiting Windows Update, I have to press "I agree" to the ActiveX control which gives whichever website I'm connected to the permissions required to remotely administer my company's computers and remotely access our data.
Hope everyone at Microsoft has got security clearances authorised by the country I'm working for, otherwise it might technically be illegal to run Windows Update...
While some people value uptime/availability/reliability, and others value security, neither of those groups will be running Windows. Laughing at one or the other only makes it more clear that (a) Windows will fail, (b) Windows will fail your security policy, and (c) Windows will expose your organization to unnecessary risk.
I thought it might be the start of a new campaign.
They're behind schedule for Windows Server 2003 Compute Cluster Edition.
It's been in the news over the previous few days and "rescheduled" for early 2006. Of course, they announced six months ago it would be out now.
They're just reminding everyone they're still in business.
And speaking of Microsoft vs. Linux, a guy at the local SQL Server user group meeting last night insisted it would be in Microsoft's best interest to come out with SQL Server Linux Edition.
I pointed out two things would have to happen first:
1) You'll hear someone point to the sky and say, "It's a bird! It's a plane! It's a pig!"
2) You'll see Bill and Steve walk naked, hand-in-hand, from coast-to-coast.
I'm a 14 year experienced developer and didn't graduate from college. I think the bias's lie within the people and not within the infrastructures. The true difference between a good developer and a bad developer is their dedication and not their education. I spent 3 years in college and barely learned anything new. In fact by the time I was out of college most of the crap I was taught was outdated.
I had been programming software since I was 13 years old and maybe I'm an exception, but I run into idiots that make these same stupid comments everyday. I usually end up repairing their shitty code. That's why when developing software you shouldn't jump into any bed. The hardest part for most people is to realize the right tool for the right job. I'm not saying I would deploy a windows server, but I wouldn't deploy a Linux desktop in a work environment.
... and here is the main reason why!!
..... being free to integrate via compatability or the openness to make compatable ......
proprietary software comes with a given level of incompatability and constraints as to what you can do with it.
Speed of implimentation of something new doesn't mean the users will adapt it as quick as the implimentor installs it.
On the other end there is the company politics and red tape to get approval to impliment something new and cost considerations of purchase.
On the open software comparison side, the implimentation of something new
There are two completely different methodologies here dealing with the primary subject matter of software.
Microsoft has a reputation of being faster to market because they cut corners in quality to do so.
Open Source Software is in no big rush to get anywhere, but moves forward more so because those involved genuinely want to move it forward, Someone who is getting paid to get something to market fast will not only sacrifice quality but may also get a bit blind as to what people would really want if they were allowed to know better.
There was a time when MS ignored Open Source software such as Linux. Believe it or not. But today, its a historical provable fact that they are not doing that any more. Slow and stead has a stronger and more stable foundation.
On innovation? Maybe that's one of the reasons MS is doing this! Maybe they just want open source software developer to hurry up and innovate something that the great imitatior (not innovator) will imitate but get to mass market sooner....and again claim they did it first...
Someone really should teach MS how to count..... grounded zero comes before 1st level distortion.
A. Cost of Leon setting up server with 3 services (DNS, email, fileshare): 4 hours @AUD$120
B. Cost of Leon attending on site roughly once a year: 2 hours @AUD$120
C. Cost of Leon remote-adminning random stuff 4x a year: 0.25hr @AUD$120
Nett cost of Leon over 2 years: A + 2 x B + 2 x 4 x C == 480 + 2 * 240 + 8 x 30 == AUD$1200 (+GST)
D. MandrivaClub Silver membership, per year: EUR$120 == AUD$191.54 (x 2)
TOTAL: AUD$1583.08 (+ AUD$120 in GST) or AUD$791.54 pa or AUD$39.58 per user per annum
A. Cost of random MCSE setting up server as above: 4 hours @AUD$80
B. Cost of random MCSE attending site roughly monthly: 1 hour @AUD$80
C. Cost of random MCSE remote adminning roughly fortnightly: 0.5hr @AUD$80
Nett cost of random MCSE labour so far: A + 24 x B + 52 x C == 320 + 24 x 80 + 52 x 40 == AUD$4320 (+GST)
D. Cost of Windows 2003 Server (Standard, OEM): AUD$1105 (+GST)
E. Cost of 15 extra user seats: AUD$272.80 (+GST and x 3)
Cost of OS software: AUD$1923 (+GST)
F. Cost of virus scanner: approx AUD$80 (or may be seat-bound depending on vendor)
TOTAL: AUD$6323.40 (+GST) or AUD$3161.70+GST pa or AUD$158.09 per user per annum
That's going to be significant to one figure, maybe, so call it three grand a year vs eight hundred bucks a year and less downtime. Call it four times as expensive to install and run MS-Windows.
It also uses a free 3rd-party MTA for the MS-Windows solution. If you wanted MS-Exchange, the cheapest way to get that is SBS, which nudges the software cost up by AUD$1657+GST, or more than the entire Linux software and setup cost.
Also, I'm being kind to the MS-Windows side by assuming 20 users. The 21st user costs $272.80 (or for SBS, $635) extra, nudging the total cost by a further 10%.
On top of the facilities provided by SBS, the Linux server software includes a choice of SQL databases, a choice of webservers and wide choice of scripting launguages, several complete development environments, a virus scanner (for protecting the MS-Windows clients behind it), highly advanced routing/filtering/mapping/firewalling facilities (including Layer7), a range of VPN technologies, complete thin client support, failover support and assorted clustering tools, and too many other services and packages to list.
Got time? Spend some of it coding or testing