Slashdot Mirror
Samba Beats Windows IT Week Labs Test Results
jmhowitt writes "Tests by IT Week Labs
show the latest version of the open-source Samba file and print server software is 2.5 times faster than Windows Server 2003 in the same role.
The news comes as many firms are grappling with the consequences of Microsoft ending support for NT4, coupled with uncertainty about when Microsoft will next update Windows. The performance difference between Windows Server 2003 and Samba 3 has increased dramatically compared with Samba 2 and Windows 2000 Server."
However, even if it's quicker than Windows Server 2003, NFS still seems to do a great deal better on my home network for the same things. For example, I typically get 10%-20% of the transfer with SMB as I do with NFS.
So I don't recommend using Samba at all unless you're looking for Windows compatibility.
Try not. Do or do not, there is no try.
-- Dr. Spock, stardate 2822-3.
they've got the ideas, but they don't have a clue how to apply'em in a working enviroment.
See article...
Trolling using another account since 2005.
This may not be the best marketing quote for Samba in the current light of events. People want to here security, not performance.
Apart from how bloody quick it is is the fact that you can log every transaction. This is immensely useful in a mission critical environment when you have to figure our exactly why one person in particular out of the entire network is having trouble. Check your Samba logs and 99.9% of the time your answer will be there.
As a system administrator I appreciate having that level of scrutiny on any network I take care of.
"And then I visited Wikipedia
Now where are the numbers to back it up?
Forget thrust, drag, lift and weight. Airplanes fly because of money.
Is it just me, or is there no testresults at all in this article? I mean, they could've at least made a little graph, specs of the hardware used or something...
Instant Karma's gonna get you Gonna look you right in the face -- John Lennon
It was only some days since i read a (biased) article of how much faster windows 2003 was than Samba. I wonder what version they compared against? 0.1?
Seems like samba is keeping the lead despite MS having thrown every service known to man far to close to the kernel to be comfortable.
HTTP/1.1 400
I read a while ago about some of the SAMBA developers having a better grasp of how the services / protocol all tie together, than the M$ employees doing the development. Most of the current M$ team inherited code from the older versions of the OS, and they are merely building on top of this codebase. The SAMBA team have had to reverse engineer the protocol. So it seems to make sense therefore, that should you understand it better, you can sqeeze more out of the service on the whole. It therefore appears that it can only get better and better as they develop .. .. And with it being opensource, bugs are easier to find ...
I also don't know how many developers are on the samba team in total (contributors / developers), but I would almost start assuming more than the manpower assigned by M$ to this area of code for Windows
... with Panther. Longhorn (or some interim SP) should include Samba 3 for CIFS protocol support... *evil grin*
Mi domando chi à il mandante di tutte le cazzate che faccio - Altan
Microsoft is good at comparing their latest product with yesterday's news. The Microsoft figures were a comparison with Samba 2. Not only that, the Samba software configuration was not exactly great if memory serves me well.
Where are the numbers?
Where are the graphs?
The article basically quotes some guy (who is actually selling Samba and thus has a vested interest) saying that Samba is 2.5 times faster than Windows 2003.
Now I have no reason not to believe him, but I was expecting a little more. And I'd wager the suits considering switching to Samba also expect more.
Is it me or does the ACL situation of *nix just suck compared to NTFS? Do I have to go commercial? Maybe Novell-on-Linux (free personal edition?)? Or are decent* ACLs in the works?
Doesn't really matter if Samba is the most badass file server ever, SMB or otherwise, if managing TBs of data is like pulling teeth, departments will choose NT.
*I know there are ACL filesystems for Linux, but from what I've seen they're pretty basic.
Opportunity knocks. Karma hunts you down.
A frequent rule in the Windows business is to split systems up over many machines. Which is great for Microsoft because essentially, you pay per client connection. With Linux/Samba, you pay according to the support that you want.
The really good thing in 3.0 was allowing the participation in ADS networks (Win 2K) as well as NT4.0. Domain controller support could be better for ADS, but otherwise it is fine.
See my journal, I write things there
That the CLIENT version of Panther is a better Windows file server than WinServer 2003... at a fraction of the price.
Too bad Macs are SO expensive. An eMac and a big FireWire array might set you back a whole 1500 bucks or so.
{/sarcasm}
.. to avoid the inevitable slew of 'but how fast does it run Quake?' gags. No doubt they'll be released later when Nvidia - sorry, I mean Samba will be accused of tweaking their software so as to get unfair benchmarking results.
Read the original poster's text again. Amazing how if this text were written comparing MS to Samba in the *other* direction everyone would be up in arms about the FUD value!
We need to be careful that we don't end up tarred with our own brush!
A little planning goes a long way...
I appreciate Samba, especially with the PDC stuff that obviates the need for costly NT server licenses here at the $workplace. Great to see that a hack that was born out of need is running circles around the cream of the Borg's crop.
Also, I agree with the rest that I'd love to see the numbers to back up the claims. Not that it really matters though. With samba you get a real good solution for an infintessimal fraction of the price of the Microsoft malware :)
We all know anytime someone publishes a benchmark favouring Windows (and there have been quite a few - tpc.org being a great example), it is instantly ripped to shreds, so why is this different?
We all know that it's impossible to do a benchmark that all parties think is fair and accurate.
Read reviews of shopping cart software
I serve printers from samba boxes to WinXP and W2k clients. I do not like dealing with setting up print queues on unix (unix printing and modem handling are evil, created by spawns of satan to make systems administrators miserable for all eternity), and I don't like Samba's way of dealing with them. It's still a bit too black-magic-swing-a-cat-over-your-head-at-midnigh
I've found CUPS to be a magnificent way of dealing with this; the combination of Samba, Unix, and WinXP/2k actually deals with printers very nicely over IPP.
Cole's Law: Thinly sliced cabbage
If the Samba team is this good, Microsoft should hire them, fire their existing developers, and port Samba to windows.
Interesting. Finally I have some hard evidence for clients who keep wasting my time asking me to support SMB on my network. I've always been telling them to RTFM and just use scp like everyone else here, but they always cry and moan that they don't know how... One has to wonder which part of RTFM don't those morons understand... But anyway, it's good to know that my intuition to avoid Microsoft protocols like a plague turned out to be right as usual. Recently all of them ask for Samba more then ever before and I think I may finally set up NFS for them one day. I just have one more question. What are the differences between Samba and NFS security-wise? I need one more argument to my arsenal.
Personally I don't give a rat, but all of those people... You know the type. They won't touch anything which is not sacred by their good uncle Bill... I need as many hard arguments as possible to prove my point here. Thanks.
Sincerely,
Pan Tarhei Hosé, PhD.
"Homo sum et cogito ergo odi profanum vulgus et libido."
Dunno if anyone else noticed, but when I clicked on the article, a "VNUNet Special" opened in the background, which was an advertisement or promotion under another name. It was formatted just like all other VNUNet articles, but was clearly a Microsoft sales pitch for W2003, complete with a flash advert on the right, and one at the top, both for W2003.
Interestingly unbiased, when clicking on a Samba article...
jer
We may be human, but we're still animals
- Steve Vai
There are a lot of people asking about the numbers.
Well, I get the print edition of IT Week and the numbers are there on page 19, in the form of a lovely little graph. The (print) article says they used a HP ProLiant BL10 eCLass Server (900MHz PIII, 40Gb ATA, 512Mb Ram) and goes in to a little detail about the benchmarking software used.
I couldn't see a copy of the article on their website but you can download an electronic copy, in some god-forsaken windows only ebook format, from www.itweek.co.uk/ebook.
Boring Old Fart (40, married, 3 kids...er no...make that 49, married, 3 grown up kids...it's been a long time)
Why doesn't everybody use SHFS (http://shfs.sourceforge.net/)? It allows for secure mounting through ssh? No need for NFS or Samba any more...
If you want to copy the best (IMHO) ACL filesystem, copy Novell NetWare. It rocks.
About a year or so ago, Novell put up on their website a script to create a large directory tree. You use this script to create these directories on both NetWare & Windoze servers. Then you start playing with the ACLs on the two trees.
In NetWare, the ACL changes happened almost instantly.
They stopped timing Windoze after 24 hours....
NetWare might not be free, but it's filesystem is the one to beat.
I know this is more of a AskSlashdot question...
My impression of Linux/Unix systems has always been that each host has it's own set of user accounts and if I have 3 hosts it means that I have to maintain 3 sets of passwords. With NT4/Win2000, my servers share a common userspace so that you only have to maintain a single user account. Is there something under Linux/Unix that does this?
How easy is it to drop a Samba server into an existing Win2000 network? Our Novell 5 server is starting to show it's age (file/printing only) and I'm starting to wonder whether to move to a later version of Novell, switch to Linux/Samba, use a NAS device, or just load up another Win2000 server.
(With the security issues this year with Windows, however, I'm not sure I want to make Windows our main file server.)
Wolde you bothe eate your cake, and have your cake?
I am a big fan of OSS, *BSD, Linux, Samba, you name it, but I have to call bulllshit on this one. A 2.5 x speed advantage can only mean one thing: misconfigured Windows machines. In my own experience (running a FreeBSD box with Samba 2.2.8 on a LAN full of Win2000/XP boxes), the Samba speed advantage has been about 30-40%.
Where are the system specs, charts, graphs, actual numbers ?
My impression of Linux/Unix systems has always been that each host has it's own set of user accounts and if I have 3 hosts it means that I have to maintain 3 sets of passwords. With NT4/Win2000, my servers share a common userspace so that you only have to maintain a single user account.
...
...).
And you would have a similar impression if you only deployed individual Windows NT/2k servers
Is there something under Linux/Unix that does this?
Unix typcially uses NIS, NIS+ or LDAP, however samba also provides Winbind for using groups and users from a Windows domain.
Plus, samba3 can use LDAP for storing it's account details, making LDAP the best choice for enterprise account management (if anyone was thinking about using NIS
The mandrakesecure.net site has some good articles on setting LDAP and samba (2.2.x with ldap support compiled in) up as a single authentication source.
How easy is it to drop a Samba server into an existing Win2000 network?
With samba3, trivial. With samba-2.2.x, you had to set certain options on the win2k domain controller (due to samba-2.2.x not having AD support).
Use AFS, then come back and say that the proprietary Netware system is the one to beat. :-)
May we never see th
This is Samba 3.0 and has huge performance benefits when running in high load enverioment.
At home when only one or two client connects at the same time, i don't get any performance difference.
But nfs and ftp utilises the bandwith better, so when I do big file transfer I use that.
...is why the implementation of Windows Server 2003 and Samba aren't more similar in performance.
Even *if* Microsoft had any respect for GPL'd code, they could have had a team to look at the Samba sources and make notes of it and give the notes to the developers, and let the developers legally figure out how Samba implements the protocols (without ever seeing the source).
I'm a networking, sysadmin, programmer (mostly programmer) consultant for small businesses in Sarasota, Florida. Most of my customers are small businesses (less than 12 people) and are looking for ways to keep costs down.
After proposing a new 2.4GHz server with Win2k3, they were sticker shocked and decided to not hire me for the job. Then one of THEM mentioned Linux (which I love and hav used for 5 years). I told them that I use Linux in my software development practice, and we could consider this as an alternative for File Server (Samba), centralized security (ldap) and backups (Mandrake backup utility). We're also using VNC (realvnc.org) for remote desktop. I can also easily SSH and do remote X session from my office, or use VNC.
It's been up for a week now, and they LOVE IT! It's fast, flexible, and you cant beat the price. And I've learned my Lesson to be mention Linux even when they specifically ask for Windows (I'm not a pushy sales person, but I do believe an presenting choices to my customers)
They wanted to outsource their IT department (the owner doesn't ever want to worry or think about their IT issues), so we made a deal that allows me to keep their systems updated, but doesn't force him to hire an on-site IT person.
Speed was NOT an issue for the Samba server, since they mostly use MS Office (win xp pro workstations) documents. However, this was a great step for them to embrace and support open source software (I donate to several projects in turn).
I hope this story might help somebody who is considering doing something similar. I'm happy to answer any questions about our experiences.
-Scott James
Suncoast Linux - Sarasota, FL
As the link seems to go up and down like a see-saw...
(Dang slashdot effect)
Here it is:
Tests by IT Week Labs show the latest version of the open-source Samba file and print server software is 2.5 times faster than Windows Server 2003 in the same role.
The news comes as many firms are grappling with the consequences of Microsoft ending support for NT4, coupled with uncertainty about when Microsoft will next update Windows. The performance difference between Windows Server 2003 and Samba 3 has increased dramatically compared with Samba 2 and Windows 2000 Server.
Samba provides file and print services to Windows PCs. It enables a Linux or Unix server to work as a file server for client PCs running Windows software. The new version also introduces support for Microsoft's Active Directory for allocating and controlling user access rights.
Samba 3's scalability is as impressive as its performance. While Windows performance scales up well initially, it then drops off quickly as more clients access the server. In contrast, Samba 3 offered excellent throughput up to the limit of our test, conducted on a low-spec Intel server.
Consequently, firms could save money on licence fees, and on hardware, by using Samba instead of Windows servers. IT consultant Alan Munday, who recommends Samba to his clients, said the resulting saving was a key consideration for firms.
Munday is already using version 3 at some smaller sites. However, he was dismissive of Active Directory. "I'm not [interested in the new Active Directory features] because the primary focus for my smaller clients is to migrate away from their ageing NT servers."
Just have each vendor send a group to configure the systems. Give them them both the same amount of money to purchase the systems. Once the benchmark has been run, have the other teams point out the flaws in the setup in each system.
Samba 3 is a very different beast from 2.2, your remark is equivalent to basing an opinion of Windows 2K on your experiences with DOS 5.0.
The numbers are in the dead-tree edition, I'm told. I don't know if they actually show any real information, because I haven't seen them.
Samba had a 2x speed advantage over Windows NT 3.51 when that was the current MS offering, though, so I don't find this completely unbelievable.
My impression of Linux/Unix systems has always been that each host has it's own set of user accounts and if I have 3 hosts it means that I have to maintain 3 sets of passwords. With NT4/Win2000, my servers share a common userspace so that you only have to maintain a single user account. Is there something under Linux/Unix that does this?
Of course there is a way. You can keep your users, groups, and passwords defined in your Windows domain controller and have Samba on the Linux box set up to use "winbind" and authenticate users against the list stored in Windows-land. That way you need no more than root and a couple daemon accounts set up on the Linux box. I have a big Linux Samba fileserver set up with a couple hundred gigabytes of autocad drawings and digitized map files, because it is twice as fast as Win NT4, but the rest of my network is NT4 and all my user logins are kept in Windows. It took about a day to set up and debug, but the Samba machine is now just a "domain member" server of my Windows domain, and I even have ACL support set up since the shared filesystems on the Samba box are XFS with acls.
Its all good this being reported to a website where everyone knows of such open-source software and its benefits. But, as in my company (and I suspect the vast majority of others) the people in charge of that sort of thing have never heard of Samba and such benefits. It is these people in these companies who need to know about it or else they will keep on forking out the millions for the slower and error prone but point-and-click-happy windows software largely because many of them are unaware of suitable alternatives!
Expect a new series of ``independent'' studies showing that Win2003 is somehow superior to Linux+Samba in terms of TCO. Of course, they will only be able to show this by slapping an Oracle database on the file server in order to skew the costs in Microsoft's favor.
CUR ALLOC 20195.....5804M
If security is your worry, use ssh on a reasonable OS in any size environment. As the orignial poster said, Samba is only useful when you have brain dead M$ client machines. If you have a real OS on the desktop, you don't need M$ protocals. Samba, as good as it is, implements M$ holes, so that M$ transmitted diseases from your client boxes can fill up or wipe out your shares after calling home and giving away everything you care to keep to yourself. Security fails with the weakest link and that will be those nasty old M$ PCs as the Half Life people recently discovered.
Real agencies worried about security have gotten away from Microsoft. I spoke with a Federal Employee last week who told me about her locked down Linux laptop. It did what she needed it to do. Real information management comes with real hardware and software ownership. Real software ownership only comes through free software. If you are running M$, someone else owns your hardware and your data and you agreed to it with the EULA.
DMCA, Hollings, Palladium. What might have sounded like paranoia is now common sense.
Please explain what this could possibly have to do with finding a superior implementation of CIFS/SMB? Novell's implementation, to my knowledge, is SAMBA, and NetWare still runs much better (performance-wise) with the client, which defeats the entire purpose of finding an implementation that doesn't require special client software (at least for Windows).
Also, considering the large base of Linux users here and Novell's lack of a supported, Novell-developed Linux client, NetWare makes a poor choice, whether you're looking for high-performance, low configuration access (on the client side) or good Linux support.
Given the assumed items of importance in this article, Windows would seem a better answer than NetWare if you want ACLs and low-configuration on the Windows client side.
After proposing a new 2.4GHz server with Win2k3, they were sticker shocked and decided to not hire me for the job.
You know, I am one of the biggest MS-bashers on the planet, and love to replace Windows with Linux everywhere I can, but buying Windows 2003 server and CALs is not really all *that* expensive for a business that depends on a network.
Windows 2003 Server Standard Edition with 5 CALs retails at CDW for about $950. Additional CALs are about $35 each. For a ten-user network that adds up to total MS licensing cost of a bit over $1100. that's only about the average price of one single typical mid-grade, name-brand office desktop PC with monitor from Dell/Gateway/HP. Not that big a deal for a business.
Because the reviewer is not taking a percentage of a non-extant samba licensing fee, that's why. It's much easier to trust free software reviews than it is to trust people trying to sell you binaries. People in the free software world have many options to chose from, why would they lie about any of them? Think about it and ask again if you are still confused.
DMCA, Hollings, Palladium. What might have sounded like paranoia is now common sense.
Even assuming this is true, does it matter?
I mean, has it ever actually happened that somebody says "Gosh, my 2Ghz fileserver with fast ethernet and half-a-gig of RAM serves files too slowly?"
I don't think in the last 5 years I've seen a fileserver running above 5% utilization unless the virus-scanner was doing a sweep.
I'm using Samba with JFS filesystems on SuSE 8.2 and have ACLs that use the Windows users/groups from the NT4 domain controller just fine.
and the results show conclusively that 2003 Server admins smell much better than Linux admins
this is due to the large amounts of organic matter that linux admins tend to eat. Windows 2003 admins only eat what is handed to them from the PHB's and eat it in the way they are told to (good puppys) and therefore are also required to wear a product called "old spice" by corperate mandate.
Gee, with $40 billion in the bank, MS can't compete with something that's free.
And that's not a surprise because all markets tend to becoming commodities over time. And FOSS is the commoditization of software. It's not a question of if proprietary operating systems dying, it's a question of when.
All you Microsoftista MCSE's can put that in your pipe an smoke it along with the crack.
From the article, the first impression one would get is that "Samba" is an operating system. After all, they're comparing it to Windows Server 2003.
No, wait, it's not, it's a service that runs on top of an operating system.
So which OS was used in the tests? Huh, it doesn't say.
This article is worthless.
...and possibly a Windoze box next to every Server.
As it happened in another 'independent' study, that stated Windoze is cheaper since you'll have to add the cost for a DOS box (Hard- and Software) to your Linux desktop to run M$ Office.
my 2 cents
For 99% of us though, that is extremely impractical. One might as well not read hardware reviews at all, and rely entirely on self-experience. Compromises must be made in whom and how we trust. I've found that benchmarks that I do not confirm myself are just about always accurate (after purchasing the actual hardware and testing it myself) when they are replicated by multiple sources.
The original thread author mentioned about using NTFS under Linux because of the poor ACL support available in Linux.
I recently got a consulting gig to migrate a large (300+ computers) Novell Netware network to Windows 2000 (or 2003) acording to the wishes of the management. And let me tell you....for properlly administered large networks Netware is hard to beat. Sure the blank screen that a Netware server presents if more intimidating than even a bash prompt. But the flexibility of NDS is awesome. Everything seems intuitive. ActiveDirectory!? That's only useful for delegating administrative priviliges. Everything else is just like a flat domain. You can't even create users with the same username in different contexts. As for file system ACLs? Same thing...everything flows downward, you can assign, filter, etc... and it's quick. Filesystem access, native file services for Netware clients, Windows (or SAMBA) clients, and NFS clients. Everything coordinated from a central directory service. And don't bring LDAP into this discussion, I'm talking a solution that works out of the box. It's not free, it's expensive...but for large networks it's worth it.
Ohhh....and Novell's Linux strategy is MUCH friendlier than Microsoft's. Soon we'll have Netware services running in Linux!!!
Luckily it seems we'll be sticking Netware for a while.
please excuse my apathy
To make up for all the lost time trying to get it working :)
Seriously though it can be rather fiddly to get it all configured right, but at least having all the settings in one config file lets you back it up once you really do get that perfect setup.
MicroSoft has a history of maintaining its monopoly by breaking compatibility with competitor's products by subtily changing (or they claim its extending and enhancing) the protocol. The most famous example were DrDOS and Java. If Samba gets too close, I wouldn't be suprised if MS didn't come up with an "enchancement" to Active Directory or SMB/CIFS or the NT-authentication protocols that will break Samba. The up-coming service pack will be the perfect oportunity for a "security fix" that will wall out Samba for a while.
(Related but slightly off-topic) A few days ago, there was an article about IE having broken support for standards, especailly CSS. I don't think that is an acident. I strongly suspect that MS won't fix IE because the "problem" helps them maintain a monopoly in browsers. If you want to get your stuff to render properly in 95% of people's browsers, you have to code to IE, not the "standard". This means your stuff won't render properly in the other 5% of browsers unless you go through lots of trouble to do browser dectection, alternate pages, or take lots of care for cross-browser compatibility.
We, as techies/geeks/nerds/zelots , need to step back and take a look at this article. While I am glad to see that someone has said that SAMBA is faster than win2k3, there is no proof. We need to make sure that we dont become like the ones we usually bash and send out and praise biased testing without real world benchmarks with published results. All this article looks like when I read it was FUD. It gave no facts, but makes a wild claim. It is possible, but wild none the less. Careful, if the OSS community keeps putting out this stuff as fact without proof, we my just get hired to replace Darl when he gets fired.
Stop signs are only Suggestions
Here's my independant study, which is not biased in anyway towards either system.
Here are my test systems and benchmark results:
System 1 - Windows 2003 Server 3.2 GHz Pentium 4 with 1 GB of RAM and a RAID array of SCSI drives 100Mbit Ethernet connection.
System 2 - Windows XP Professional 2.4 GHz Pentium 4 with 512 MB of RAM and a 7200 RPM Serial-ATA drive 100Mbit Ethernet connection.
Here are my results of downloading a 500 MB file from the server using CIFS:
Time it took to recognize the system: 30 seconds
Time it took to transfer the file: 3 minutes
Now, with Samba:
System 1 - Redhat Linux 9 with a Pentium Pro and 32 MB of RAM 10Mbit ethernet connection
System 2 - Redhat Linux 9 with an AMD K6-2 200 MHz, 48 MB of RAM, and a 10 Mbit ethernet connection
Here are my results of downloading a 500 MB file from the server using CIFS:
Time it took to recognize the system: 3 minutes
Time it took to transfer the file: 2 and a half hours
Now there's proof that Windows Sever 2003 is significantly faster than Redhat Linux w/ Samba.
Hypocrisy is the 8th deadly sin.
Quite an insightful comment I must say, at least for an argumentum ad personam. This was probably one of those "trollings" I'm hearing so much about, but still I will answer.
I was talking about security, not convenience. By security I mean the cryptography and number theory behind the protocol, not a "wide selection of permissions."
The rest of your long and boring comment is a childish and baseless attack on my person, so please excuse me if I ignore it before you start comparing our penii length.
Sincerely,
Pan Tarhei Hosé, PhD.
"Homo sum et cogito ergo odi profanum vulgus et libido."
...when I tell you that my daughter's tricycle is 2.5 times faster in the quarter mile than a Ferrari F-50. ;)
ABSURDITY, n.: A statement or belief manifestly inconsistent with one's own opinion.
Samba provides file and print services to Windows PCs. It enables a Linux or Unix server to work as a file server for client PCs running Windows software.
Looks pretty much like they ran Samba on Linux, right? Or maybe they ran it on Solaris, but that's sorta stupid.
Not that long ago, MS hyped a "benchmark" using some old version of GNU/Linux and some old version of samba on a poorly configured Linux box and put it up against a tuned MS Windows 2003 server to show "how much faster" MS Windows 2003 is. Of course, the "benchmark" was funded by MS. Now a non-biased benchmark comes along and makes MS look just silly. Were not talking 5% - 10% faster, but 2.5 times faster! That is a big difference.
If Tyranny and Oppression come to this land,
it will be in the guise of fighting a foreign enemy. -James Madison
...if anyone's done any work on running kernel modules with user priviledges, or VFS interfacing with a userland daemon, like the way Netfilter does with ftwall.
I'd have a blast working on it. Unfortunately, working/going to school for 80 hours a week just doesn't leave much time for other stuff. I tried going without sleep, but somehow my boss noticed a difference.
Does anyone know of any statistics describing how much time a kernel developer spends on the kernel?
tasks(723) drafts(105) languages(484) examples(29106)
This wouldnt have happened to you IF you would have used Mozilla or likewise with POPUP KILLING. Let's see that one more time, with Mozilla...
;-)
jsmyth>......
See!!!! He wont say anything cause he wouldnt have seen it! Works like a charm
Though not strictly true, you mean the Samba team has refactored the SMB protocol. Microsoft on the other hand has made it a point to not touch the existing historical routines until something is found to be wrong with them. When you refactor, the chance of effecting millions of different clients that depend on the initial very strict protocol timing is critical. Though Samba is faster, is it more reliable? Even more so, since they've not only refactored, but they've reverse engineered also? There is nothing wrong with what they've done as long as the finite state analysis reduces correctly.
I for one applaud this kind of thing because I hate the fact that very few software programs get optimized over time. As a systems programmer, I resent moving on to other projects until I'm sure I've got the absolute fastest code. But alas, time, money, and resources ultimately give way to the need for change. It really just sucks. Way to go Samba team!
-1
I have been trying for the past month to get 3.0.0 to work with a Windows 2000 PDC, and still can't get ir right. I've gotten everyting BUT Samba to do it, (actually my goal is to authorize users from Active Directory to access shares and what-not). Part of the reason Linux is still hurting is beucase you have to jump through hoops to get this stuff running. I spent a whole day jacking around getting dependinces going for this and it really pist me off that Samba didn't list them. My vote -- Linux still ain't ready. Too much work to do the easiest things (or at least should be easy). I have yet to find a HOWTO that was up to date and actually worked for what I needed it to do... Linux documentation sucks. Yes, this may sound like flaimbait (and it's not), but how much easier is it for Windows to get things going before Linux? I can get my Win2K PDC up faster and more secure than I can my Linux box because of problems like these. (Go figure, this issue happened recently to me... so yeah) What will really impress me, is when a distro can do this out of the box. Communicate to Linux and Windows securely (without a registry hack on the Windows side... I don't play plaintext security.. it's like going without a condom). This will allow admins to 'try' linux for their bosses and allow a migration over. Some small buisnesses can't afford a sink or swim solution. My $0.02
"Do or do not. There is no try." -- Master Yoda (Half man, half muppet)
Windows 2k, running the Java version of Samba, Huhahahahahhaahha
Wow, I should not post when knackered.
Number of clients (more == better)
Raw Data:
Win 2003 Server: ### (3)
Samba Server 3 : ############ (12)
Normalised:
Win 2003 Server: # (1)
Samba Server 3 : #### (4)
As visible Samba is able to handle 4 times the number of clients before performance takes a hit.
Yet Socrates himself is particularly missed.
A lovely little thinker but a bugger when he's pissed.
No, SMB is and always has been a very published spec.
l
See this informative page:
http://samba.anu.edu.au/cifs/docs/what-is-smb.htm
I browse at +5 Flamebait- moderation for all or moderation for none.
For fuck's sake- it's a very valid point...
I browse at +5 Flamebait- moderation for all or moderation for none.
There is a scant amount of information on the actual tests performed in this article.
- "History shows again and again how nature points out the folly of men" -- Blue Oyster Cult, 'Godzilla'
Ohh, but if I wanted to use Samba on Panther Server I would have to un-rack one of my Dell 6450 servers and fill the hole with six Apple 1U servers.
For about what the Dell cost.
And I'd have to stop using Nt4 or 5 or Wind0ze server that I paid extra for, and use the OS that ships with the Apple boxes.
butButBUT!!! I paid a LOT of money for that software!!
Oh, I know, I won't have to throw away the Dell and it's virus-laden OS because my controller will not let me buy six CAT5E cables and a new Gigabit Ethernet router to get about 20 times the performance per U rack space.
"Too much money."
====================
The Micro$oft apologists on here are all too stupid to realize that "Battlestar Galactica" was actually an allegory of today's computing world that was 20 years too early. The Cylons are M$ and the rebels are [ Mac | Linux ]
And you think for a second Microsoft would go for that?
That would mean having to compete on the merits of their product instead of relying on their (excellent) marketing department. This would give them a serious disadvantage.
Never gonna happen.
I would be interested to see a NetWare 6.5 server with NSS compete against these figures, after all if you are talking file and print performance you are talking NetWare and we all know it. As a sys admin I've stood at windows consoles waiting for file attribute information to be stamped, I've sworn at *nix for it's poor administration of file system rights but (with the odd exception in the 5.x days of NSS) I've not complained about NetWare. Novell has always had a team of dedicated programmers working on the file system and speed has always been a main concern with NetWare (in some cases far too much so) - you don't need Client32 anymore and my own benchmarking with CIFS in NetWare 6.5 shows considerable improvment over Win2k or 2k3. Put simply, NetWare as a file share platform in a large environment cannot be beaten. In a small environment I'd wonder why you'd even be interested in performance differences as you'd never see the effect. It should also be noted that NetWare 6.5 now gives 5 licenses for free. If you are going to talk about Printing I'd suggest that you look at iPrint, using the IPP protocol over HTTP you can bring up a floor-map based map of your printers for your users to click on to install on their desktops. Your average NetWare server can handle over one THOUSAND printers, novell actually states that 2000 is the limit: I've personally seen 1200 on one NetWare server.... someone show me a x86 *nix or M$ server that can have 2000 active printers please? Then there is the administration... web based via iManager 2.0, it's a dream compared to the MMC. Of course there are down sides to NetWare, it's not cheap (again, who cares we are talking performance and therefore large scale businesses that can afford it) and you have to get out of your comfort zone, however this whole discussion is based on 'horses for courses' and immature operating systems need not apply.
does this mean that samba 3 is that much faster or that win2k3 is that much slower at the task?
Large print giveth, and the small print taketh away
After all, who needs incredible high user count scalability for only simple file sharing on aging hardware? And what the heck was the setup? Anything setup slightly incorrectly can suffer incredibly major performance hits.
...benchmark[s] favouring Windows [...] [are] instantly ripped to shreads, so why is this different?
It's not different: the thread is FULL of people (including you) asking "where's the numbers" and calling this study FUD. Some of the highest moderated comments, in fact.
Unfortunately, the response seems to be: MS doesn't allow anyone to publish their numbers. IIRC, they added this clause to their licenses after Oracle published some unfavorable-to-MS benchmarks.
The real difference is that when OSS loses a benchmark test, the general reaction* is that the results are studied, and if they're legitimate (which they often are), then work begins trying to fix the problem, and the eventual result is (usually) a better system. When MS loses a benchmark test, they react by forbidding any more benchmark tests, and nothing else (usually) ever changes.
* At least, the reaction amongst people who count, which is to say, the developers, not the pro-linux trolls on slashdot.
try smell in the right place..
I fully agree with you on "weblications" growing. But while we're talking about it, can we please not use that term again? And another scary stupid buzzword I've heard too much lately: BUI. As in Browser User Interface. CLI and GUI are bearable, but BUI? Ugh. It's almost as bad as Blogs.
Just my own 2cent rant.
J
HP ProLiant BL10 e-Class Server? A File Server??? oh, puhleezz! It's equivalent to high-end laptop from two years ago. However, it's very useful for things like web servers, terminal servers i.e. scale-out applications. Go look at the product page here (hp.com) and decide for youselves. The 40GB drive is a laptop Hard drive, admittedly the faster 5400rpm variety. Not a 7200rpm desktop ATA or 15KRPM SCSI, let alone an array of them that I would expect to find in a file server.
Isn't it simply awesome that a non-windows system can interoperate and participate fully within the Windows domain of networking?
It's not like the samba team has only reverse-engineered 25% of the protocol. They literally have 100% compliance minus the Windows proprietary RPC protocol running on top of SMB. And that has been attacked too; Samba can be a primary domain controller, be part of a domain, etc.
Kudos for the samba team for a job well done.
And then Microsoft comes in and offers an independent test with these folks in the same lab, so that Linux people configure Linux machines and Windows people configure Windows machines. This happened once to Apache, it was simply ripped apart by properly configured IIS (remember post-Mindcraft face-to-face independent test?).
:0)
Alas, Linux community is not learning lessons. I'd like to see Bill Almighty issue a smackdown order now.
Of course, in order to achieve the same benchmark results as myself, you must be building a home network and completely giveup on ever getting Samba to work and just decide that an old ME laptop will have to be the print/file server instead.
You tell us that "speed was not an issue" for the customer, yet you proposed an expensive 2.4GHz server with Win2k3?! Did you not know the speed requirement at that time? You could spec a very good, reliable and decent-performing 1.8GHz (or less) Celeron machine running Win2k. A Win2k Pro license (there's nothing in Pro's EULA that prevents it from sharing files very effectively) goes for less than USD$200, so that shouldn't break the bank either.
Anyways, I'm happy that you got a customer in Linux, but it didn't sound like a comparable Microsoft solution was even proposed.
Please give me some advice.
At home, I have a very small network. It's now all Linux machines (although my workstation can dual-boot into Windows for gaming).
Currently I'm still using SAMBA on my server, but I have been wondering if there is something better I could use, now that the network is all Linux.
What would you recommend? Is NFS horribly insecure? Is there anything else, such as CODA, that you would suggest? Is SAMBA really my best bet?
P.S. The number one feature I wish for on my server is file backups, like the Norton Protected Recycle Bin features you get on Windows when you install Norton Utilities. The closest thing I have seen on Linux would be to run LVM and set up automatic snapshots. How would you suggest solving this problem?
steveha
lf(1): it's like ls(1) but sorts filenames by extension, tersely
nuff sed ;)
Are you talking about man pages? Maybe The Manual is nothing but obvious trueisms like "-gurfle invokes the gurfle feature" and mindless wave-offs like "-glork is not within the scope of this document. See The Other Fucking Manual for details."
Sometimes users just need a handful of simple straight-forward plain vanilla examples printed out on an instruction sheet. But then sometimes a sysadm couldn't be Fucking Bothered to do something like that. That's why I get to put "good communication and inter-personal skills" on my resume... not that it's done any fucking good so far.
My other car is a 1984 Nark Avenger.
If they know how to use Windows Explorer, they know how to use WinSCP2. Point, click, drool. Of course, you can also use it for storing stuff besides porn.
Got time? Spend some of it coding or testing
That is scary, but not as bad as clueless people who continue to believe that M$ has made changes. What's known is this, M$ == always insecure.
DMCA, Hollings, Palladium. What might have sounded like paranoia is now common sense.
Microsoft likes to run benchmarks where the disk array is RAID 0.
It does give better performance under some conditions, but it is the worst configuration with regards to your data. If any disk dies, you've lost all your data.
A RAID 5 array would be more secure, but would not give as good performance in all circumstances.
If the hardware was not specified, then the platform that used the least resources would have an advantage. Instead of paying extra for the extra megahertz, redundant components could be included.
Then, after benchmarking each system, the benchmarks can be run a second time, with each team demonstrating how their configuration has advantages.
The easiest one would be the RAID 0 vs 5 example. The RAID 5 team could cut one of the power cables to a drive on the RAID 0 team's server. That server would crash and everything would be lost.
Then the RAID 0 team could cut a power cable to one of the RAID 5 team's drives. But their server would continue to run and no data would be lost.
That is how to run a benchmark of two different software platforms.
Comparing Apache on Linux to IIS on Windows. Do you go with one huge and expensive server with multiple processors and multiple network cards or do you go with a few cheaper but less powerful machines with only one processor and network card each?
Which configuration would give you the best performance in that circumstance?
Which configuration would be the most fail safe?
Are you trading performance for reliability?
*Their* implementation of tar would be closed source, and protected by a EULA. Try getting out of THAT tarball!
"No problem. I have the capacity to do infinite work so long as you don't mind that my quality approaches zero."-Dilbert
True, Novell does not currently have a supported Novell-developed Linux client. But they now have Ximian.
Those interested in the details of IT Week's test can see a graph of our findings here.
IT Week has added further information here, to meet all the calls for more data.
"The "losing" group invariably complains that the OSs weren't tuned effectively, set up improperly,etc., and then cry foul about the setup, attempting to declare the results biased and moot."
Whereas I clearly stated that each group would do their own tuning. How can one side complain that their platform wasn't tuned correctly when they did the tuning themselves?
"For example...with regards to performance and reliability...I, for one, have never (and will never) rely on raid for webservers....I'll have redundancy set up elsewhere (i.e. load balanced)."
Which was covered in my second example. Given a set budget, how would each team configure their website? What are the performance benefits of each approach and what are the drawbacks?
If you go with one server that eats up all of your budget, then running anything less than redundant disks is idiotic.
If you run multiple servers in a cluster or load balanced configuration then the redundancy is provided by the servers themselves.
I believe I had spelled that out clearly enough for anyone with any knowledge of the subject to understand.
SFTP
An earlier poster alluded to ftp's strength - it's faster than snot. And while I'm sure ftp is faster than SMB, I'd guess it's faster than NFS as well. [Didn't Van Johnson (UCB) do early speed tests for Ethernet using ftp?]
I haven't used sftp.
I'm curious how it performs relative to ftp and to scp (as well as NFS and SMB) for different frame sizes.
"Provided by the management for your protection."
"Uh. You're not understanding me. In past samba/filesharing/webserver/database benchmarks, there have been many times where the 'losing' group complained that the winning group had tuned their setup to be unfair and not within the boundaries of the test. Maybe you haven't seen reports of that happen...sorry if I assumed you had."
Did you miss the part about me saying how the benchmarks should be run? Should be run? SHOULD BE RUN?
Instead, you're going on about how OTHER tests had problems.
Try addressing what I've said instead of other situations, okay?
"The only reliable benchmarks are those performed by *real* non-biased parties, and the exact specifics of tuning, setup, procedure, etc. have been documented."
No they aren't. Because even those non-biased parties might tune one system for better performance than the other.
That is why each vendor sends out their own teams.
If anyone wants to argue that Team Samba doesn't know how to tune a Samba server, that's their problem.
It might not be a "benchmark" in the sense you'd like it to be, but it would be a reliable comparision of two platforms. Which is what "benchmarks" are.
"Your suggestion for a 'benchmark' isn't measuring the performance of anything, except for the intelligence of the people running the tests."
Nope. I can get the smartest people to do the configuration, but if the software can't handle the load, it can't handle the load.
How well would a web server from back in 1989 compare to a web server today? It doesn't matter how smart the people configuring it are, it will suck compared to a current server.
"It's obvious that most situations will go with your latter option (multiple servers, load balanced) for many reasons, not just budgetary, performance, and reliability."
Most, but not all. Remember, there was a Scandinavian ISP named Telia that deployed Linux on a mainframe to consolidate their websites. They went from 70 machines down to 1.
Different scenarios can have completely different solutions.
That is why each team should be given a budget and a scenario. Then let them do the best they can with their platform and that budget.
Saying you need to do a re-write for maintainability never gets the rewrite done. The only way is to say "Bu making some significant modifications, we can add x,y, and z.
If x,y, and z are marketable enough, the project will be a go. Your first internal milestone is to have the code surviving the old test suite with the rewrite. The first milestone the PHB sees is a simulation of x with limitations. The second is a full implementation of x. Then comes y, etc.
Well... No.
Sincerely,
Pan Tarhei Hosé, PhD.
"Homo sum et cogito ergo odi profanum vulgus et libido."