Slashdot Mirror
Unix Vendors Get Creative Against Windows & Linux
coondoggie writes "As x86 servers become increasingly capable, IT managers are taking a closer look at their Unix installations to determine whether a move to Linux or Windows might make sense, analysts say. "The defensible hill for Unix is the big, vertically scaling, mission-critical application, which is usually some type of database serving," says Andrew Butler, a vice president and distinguished analyst at Gartner. "But increasingly, the appeal of Windows- and Linux-based systems running on cheaper, commodity hardware is becoming more and more compelling.""
Linux beats it in hardware support, but Sun has the whole overpriced reliability image which some might find attractive. If you're paying the big bucks you can get a good response from Sun, though I'd suspect people working on Linux could make those bucks go further.
I've worked on some huge Unix systems (mostly for databases) and never once did anyone mention Windows without laughing. No way are people with truely large-scale critical Unix servers considering switching to Windows. When you already own the hardware, paid for the software, and have huge support contracts, consider expansion with Linux. Windows is only intruding on the smaller scale Unix installations.
Gartner is known for sometimes putting out some fluff but this just sounds silly.
Developers: We can use your help.
Last time I checked, both BSD and Solaris (which are UNIX not Linux) run just fine on commodity x86/64 hardware. Sounds like somebody missed everything from 1999 on.
Cheers, -b.
When I think of Sun, I think of reliable, mission critical, just like the article says...Sun has this big business image that "if you want it to run, you should be using Sun", but it also comes with a steeper learning curve. Whereas Linux's image is building and linux has an attitude like "anything you can do, i can do better, and if i can't yet, i will soon" and also comes with less of a learning curve...however still a lot more of a curve than your run of the mill windows server guy would like, I've met so many bleeding heart MS guys that would use/try Linux if they didn't have a misconception that it is infinatley harder than windows...
In a world of acronyms, the words are the real victims.
"Anal" - backside
"Yst" - ancient Greek word, meaning "to pull ideas from"
the island defended is getting smaller. In my experience databases are not growing in a similar speed like hardware is anymore
My experience contradicts this. Companies are analyzing more and more information and using larger data warehouses. Where in the past I'd see a variety of small databases spread throughout financial firms I now see more cosolidation into data warehouses. It aids in analysis, cuts some costs, and increases security. Even many web sites are now growing huge databases.
Developers: We can use your help.
A place I used to work planned to move everyone from sun workstations to windows boxes as a cost-cutting exercise.
It ended up costing way more overall because all of a sudden our IT department went from a single sysadmin who was hardly ever busy, because everything just worked, to a whole department of IT staff needed to second-guess MS exchange and a now very unreliable network (even though no network hardware or configuration had changed), and Windows PC's that were always slowing up or crashing, especially after that stupid automated windows update.
The big advantage that BSD and Linux have over Windows in this space (migration away from commercial Unix) is that most of what you have already learned is, as near as damn it is to swearing, directly portable. Even most of the applications you are already running, need no more than a swift recompile.
Je fume. Tu fumes. Nous fûmes!
I ordered them the day I read the slashdot article. I still haven't received them.
sadly, no... I dont expect them for a while either. The ubuntu cd's I got from them took about 6 weeks...
--MaxPowerDJ
X windows is 23 years old. UNIX, with trusty system calls like open() is around 38. Without radical innovation, its no wonder that customers are moving to low cost alternatives that coincidentally do open() or X-Windows just fine. If Sun wants its market back, they should have photorealistic 3D graphics, real time, robotics control, neural network security system, files presented as memory mapped data structures of type-specific format... There are opportunities, market and technologies that are still left for $1M price tag of high end Sun servers or Cray supercomputers. Its just that these companies have been overrun by management that has too much money and too little brains to care.
RELIABILITY!
Cheaper, commodity hardware does not work for all of us! There's a huge category they're missing out on.I know the Fortune 100 (or maybe 500?) companies don't care, as they can just run clusters of cheap ass machines. But what about the millions of small to middle sized businesses and research institutions?
I've been involved in a number of smaller sized research organizations, and uptime is the utmost importance, however, we definitely aren't running "server farms", so clustering is out the window. I've relied on Sun servers running tons of GNU tools to get the job done. I think you'll find (unless you already know) there's a very large number of people doing what I'm doing. We can't rely on Dell (or even Penguin, or Monarch, or....) to deliver consistent, well thought out, easily-repairable, robust servers. Sun (and other big box makers) can! So what do I do? Run Solaris 10 (GREAT, Solid OS) and install a ton of GNU open-source tools. The result? Great open-source software, and the reliability and well thought-out hardware from Sun. It takes a bit longer to do, but the results are great.
B E A utiful.
Can somebody explain how Unix is different than Linux? Most Linux distributions are mostly POSIX, SUS and ELF compliant. Is the underlying code better somehow?
Well, it's no wonder you're having kernel panics and reboots if you're running Red Hat -- they patch the kernel to christ and their userland stuff depends on the patched kernel. Try Slackware or Debian. They both "Just Work" with a kernel built from plain old kernel.org sources.
Or ... they could move with the market, offer an x86 version of Solaris that is the same as the Sparc version, start selling AMD and Intel server and workstation solutions, and bring themselves back to profitability (something they haven't done in years).
... they did that. Much like TFA states ...
Oh wait
- Roach
linux [...] comes with less of a learning curve...however still a lot more of a curve than your run of the mill windows server guy would like, I've met so many bleeding heart MS guys that would use/try Linux if they didn't have a misconception that it is infinatley harder than windows...
Wait, what? Linux has a steeper learning curve than Windows, yet Windows admins have a "misconception" that Linux is harder for them to use?
Either it's easier to use (in which case the learning curve isn't as steep as you claim), or it's not (in which case there's no misconceptions, only reality).
the coolest club on
I beg to differ. Here where I work, our main customer database (which is OLTP, NOT a data warehouse) just crossed over the 20TB mark. Our data warehouse is up in the hundreds of TB range, and growing very quickly.
Like all things with technology, as things progress and get larger and faster, people find ways to use that new power.
There is a _large_ market for high end systems. We're talking 32+ processors and TERABYTES of RAM. Commodity hardware has increased in power and flexibility, but the big Iron has as well.
Anyone who conflates "Unix" and "large, expensive custom/proprietary hardware" in 2007 isn't worth reading. While there are indeed some Unix operating systems that only run on custom hardware produced by the same vendor, that's by no means universally true. Note especially Solaris, which runs just as well on the very same cheap and ubiquitous x86 (whether from a tier-1 vendor or homebrew) systems used by some to run Windows or GNU/Linux as it does on the big, expensive SPARC hardware that Sun and Fujitsu offer. Anyone who wants to have a meaningful conversation about the IT industry needs to start by separating the hardware options (driven mainly by economics) from the software discussion (driven mainly by technical and business factors). While there are business problems that can only be solved on high-end hardware that's often limited to a single choice of OS, those are the minority of deployments and form a distinct market from the volume space. Talking about competition between high-end and low-end solutions is pointless; either you need high-end performance, capacity, and features or you don't. If you do, you're simply not in the market for a low-end hardware platform and the OS you run will depend largely on the hardware vendor you choose. If you don't, it would be silly to spend money on high-end gear, and you'll be able to choose from among several operating systems - including those named here - based on your individual business needs and the features offered by each product. But it's a sure mark of ignorance to discuss the two as if it's all one market in which a choice of Windows/GNU/Linux/Solaris/BSD on a uniprocessor PC competes directly with HPUX on Himalaya and Solaris on Starcat. One can see why commentators are always talking about Unix's imminent demise; they fail to recognise two key aspects of the market: Unix's strong and capable presence on both low-end and high-end hardware, and the segmented nature of the server market. Not much to see here, I shouldn't think.
The sheer ease of starting a business with Linux and other open-source technologies (MySQL, PHP, ruby, etc) gives Linux the the advantage over Unix. If the company grows, we will just keep using what works.
I am currently working in a 10-man startup company that delivers employment training over the Web. Our system runs off a cluster of 3 boxes in a LAMP configuration, and we never paid a dime for server software.
Linux dedicated hosting is much cheaper than Windows dedicated hosting, and there are so many tutorials and packages out there that make it really easy to learn and deploy open-source systems.
Sun and company have started their battle way too late for anything but niche deployments -- the King of "Big Iron", IBM, long ago threw in the mainframe towel in favor of Linux.
My Dad used to run a university library, and he was always very forward - looking in terms of IT. He wanted to get a Sun server to run thin-client systems for the library patrons to use rather than having to clean the Windows systems every day, and he could not get a Sun salesperson to talk to him (this was about 12 years ago).
The main library software ran on Sun servers (that they bought through the software vendor), and he was highly impressed with the stability of the Sun boxes. He was so impressed that when the time came for PC's to be installed in the library, he wanted to put 20 thin-client terminals in that ran sessions on a second Sun server. That plan ended because he could not get Sun to talk to him -- he literally could not get the sales people there to call him back to sell him the system.
The end result was that he had to install the 20 PC's and deal with the viruses, downloaded software and other daily headaches of the Windows world and Sun lost an easy sale because they were too arrogant to care.
Sun should have been fighting way back then -- Linux is way too mature now, and way too cheap and easy to deploy. In these days of Ubuntu livecd's and Macs running on top of Linux, anyone who is not a Windows person who is interested in computing will learn Linux. Sun may have a few legacy apps, it looks like they will just be a niche player at best. Sun was legendary for their stability, but our Linux boxes have all the stability we need.
I am sure Unix will have it's niches here and there, but Linux is way too strong at this point.
No conflict in gp: he says they have a misconception that it's _much_ ("infinitely") harder to use, but that in fact Linux has a lot more of a curve than a windows server guy would like. I'd agree--many do have an overblown sense of how hard Linux is to learn/work with, but in fact it is _somewhat_ harder to learn than Windows.
FYI, OS X Leopard *is* Unix, it's been offically certified as such and will be marketed as such, unlike the previous versions which were 'Unix Like'
Wait, what? Linux has a steeper learning curve than Windows, yet Windows admins have a "misconception" that Linux is harder for them to use? Actually that is fine. Initially, linux can have a steep learning curve (first week or two) but after that it is easy. Windows admins have to misconception it starts difficult and stays difficult.
Just junk food for thought...
i would imagine that is the case in many large datacenters. to paraphrase the great philosopher jules winnfield: mission critical enterprise applications are not in the same ballpark as windows and linux on x86. it's not even the same sport.
the one big shop i worked for in central ohio used mainframes, unix, and windows. mainframes for a lot of legacy data (like stuff from the 70's and 80's) unix for all things oracle, and oddly enough, windows for the front end web hosting. so all of the customer facing web front ends were windows based, but the data itself was hosted and served up from sun gear. inside the firewall, the company itself was seriously MS centric, being used for a number of MS case study/whitepapers on replacing X and Y with stuff from microsoft. i haven't worked there in several years, but i would imagine they are still all MS until you get to the data itself.
there was also a big project in the mid to late 90's to replace the aging development workstations. the sun workstations were replaced with windows NT4 running eXceed. again, windows on the front end, but the serious work is in unix. the support costs for having a sun on the desk for development and a wintel for office and email were offset significantly by replacing both with a powerful (for that time) PC and host emulation software. in all of the "unix substitutions" that i had seen, not once was the proprietary data hosted on anything but IBM mainframes or SUN unix.
IBM will occasionally hit the IT trade rags with ads and op-ed pieces about "server sprawl"... a kind of out of control proliferation of low-end (in IBM's opinion) servers or even clusters that handle one or two tasks instead of running everything from a handful of multi-million dollar godlike servers. i would imagine that sun will eventually take the same approach.
i think that larger and more well established datacenters will always feature commercial unix at their cores thanks in large part to the investment and contracts that you mention. but i think newer companies will focus on large numbers of commodity servers (ala google). i think that is where MS and commercial unix will fight a pitched battle with linux.
sarcasm:
-noun
1. harsh or bitter derision or irony.
Ah, the young and naive. Its not that there are people "moving on" but that more people who don't know the old stuff are entering the work force. Our company used to be a unix shop, with HP Apollos, HP PA-RISCs, Suns, SGIs, and RS/6000s. Along came the late 1990s, and some new engineers were hired. They said, "hey, look at these super Dell machines we can get! They run Windows! Everyone loves Windows! We don't have to pay for OS maintenance, and they cost $10,000 less than our current workstations!" And so the Windows migration began. We have gone through three generations of Dells in that time. Sadly, most of the old unix stations were trashed before I got here. Remaining are an SGI Iris Indigo R3000, a Sun Sparcstation IPC, and a Sun Ultra 1 C3D. Although they do not see a lot of action these days, when I do need them, they work flawlessly.
In this day and age where linux is starting to become a real engineering and commercial OS, more people are saying, "we want software on linux." With Apple slowly gaining market share, people are saying "we want software on Mac OS X." Did the people in our company ever think that once we ported our software from unix to windows we would ever be faced with the possibility of going back? Probably not. However, all these years later it isn't that far fetched now.
Radical innovation is irrelevant when you need reliable systems. If you are an IT manager that believes consistent radical innovation is key to developing reliable products, you are wasting money and should be fired. It is irresponsible to assume that something new is reliable. Who cares that X is 23 years old? Apple is using X11 on BSD and they are laughing all the way to the bank.
Where have you been the past 15 years? SGI _invented_ photorealistic 3D graphics, did you not see "Jurassic Park"? Sun systems have been able to do high-performance 3D CAD for almost as long. Heck, my two year old Dell Precision 670 runs the EXACT same CAD software as my Sun Ultra 10 C3D (UGS NX5) complete with photorendering. The Sun isn't much slower. Matter of fact, when I had Vista on the 670, NX5 actually ran FASTER on the Sun. Not bad for a computer I bought for $40. Half the software we need to run on that 670 won't run on that machine whether in XPx64 or Vista. Investing in radical innovation when we needed reliability got us NOTHING but a bill from Dell for $3,600.
For robotics control, perhaps you are not familiar with the Sun SPOT. Nifty little self contained computers, with sensors, ADCs, and communications. When they hit the market, I'm considering using them to drive a UAV.
Innovation and reliability are two completely different things. If you need reliability, why does innovation matter? When you need your data, you need it, nifty features be damned. Now, I have some FORTRAN programming to do. Time to fire up the SGI.
Very informative. Good job in bringing that to light. I guess that also will settle the litigation issue between apple and the open group over the UNIX trademark, about which I've been very curious but haven't seen any developments on.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
From the article: "especially now that Web-based applications are written in operating system-agnostic languages such as Java and .Net."
.net framework, and it sure looked 100% windows to me. Perhaps the author is considering "multiple flavors of windows" to be "multiple platforms." My bias leads me to a different conclusion.
I just went to Microsoft's page for the