Windows 8 file system gets upgraded, named ReFS

Windows 8 fans will have heard about Protogon at some point. The new file system has been spotted lurking in developer builds, but details are a little on the murky side. The important facts are these: disks can only be formatted in Protogon from the command line, Windows 7 and earlier can't read it, and it shares a lot of features with NTFS.

It looks like from here on out Protogon is no more. WinUnleaked has found that latest builds have removed Protogon support and replaced it with a new file system. Be prepared to hear "ReFS" a lot from now on, as this new system is better supported than Protogon ever was. For a start, the disk formatting utility actually lists it as an option.

At the moment, Windows 8's installer doesn't play nice with ReFS. Users are told that the system can only be installed on NTFS-formatted drives. It's entirely possible that installation isn't supported because the feature is incomplete though, so at this stage it's hard to say how finished versions of Windows 8 will treat ReFS drives.

Signs are pointing to Windows 8 seeing a 2012 release. At this stage in development, ReFS could stick as a final name: it's short, simple and easy to remember. The question now is when Microsoft will release some details on the new system.

Report a problem with article
Previous Story

Review: Scrivener for Windows

Next Story

Path for iOS review (or, why Facebook should buy them now)

62 Comments

Commenting is disabled on this article.

Okay, I'm now even more convinced Protogon... sorry, I mean, "ReFS," is WinFS. ReLATIONALFS. & for the usual, "HURR DURR WINFS WASN'T A FILE SYSTEM," people, need I remind you about OFS?

MASTER260 said,
Okay, I'm now even more convinced Protogon... sorry, I mean, "ReFS," is WinFS. ReLATIONALFS. & for the usual, "HURR DURR WINFS WASN'T A FILE SYSTEM," people, need I remind you about OFS?

That's not the name of it, actually the Re in ReFS seems to stand for Resilient FS and will be a server OS feature. So in the end it'll be in Win8 Server and not on the desktop it seems.

GP007 said,

That's not the name of it, actually the Re in ReFS seems to stand for Resilient FS and will be a server OS feature. So in the end it'll be in Win8 Server and not on the desktop it seems.


I typed that before it was reported. BTW, winrumors said that, "sources familiar with the company's plans," said it, doesn't necessarily mean it's true.

EDIT: Also, as I've said before, Protogon was in leaked client builds.

It looks like there's much less metadata in "ReFS".
An empty 30GB NTFS volume has 90.6 MB of used space (I'm no expert but most of it must be the MFT).

The only thing I'm actually liking of this is the ability with the Microsoft Xbox console. For now, the 360 only spots FAT which sucks, no more than 4GB files can be read by the system.

Hope that ReFS and the new Win8 integration to their system give us the ability to actually playback so desired file sizes.

SK[ said,]Stupid water mark right across the part where it mentions ReFS.

Considering that its Microsoft intellectual property, they should not be water marking it. Steve Sinofsky should find them and delete them. The recycle bin their @$$

blade1269 said,
Dont reall care about a new file system JUSt Fix the user permissions in windows... So many errors...

OSX user? What problems do you have with Window's ACLs?

1) Why not consider that it may be the official name for the Paragon codename d FS, in a new unfinished state? (it sounded like a codename coming from MS)
Yeah that may seem like a lot of guesses but i find it farfetched that they would have developed 2 file systems in the cycle of a single OS version.
2) How come we haven't seen complaints on the old disk usage pie graphic yet?! I'm surprised.
3) In the end I'm glad that some big basics are being rethought, even if i don't agree with metro for desktop usage with or without touchscreens. Microsoft didn't use to be so daring.

My three cents

It's weird, in the GNU/Linux/FOSS world, we have so many new file systems coming out all the time, it's hard to keep up btrfs looks like it could be the next contender after ext4.

Joey S said,
It's weird, in the GNU/Linux/FOSS world, we have so many new file systems coming out all the time, it's hard to keep up btrfs looks like it could be the next contender after ext4.

Because they are free software. So they can carelessly test these kinds of things whenever they want. But Microsoft can't do it. Billion of people depend on it. Windows is not a toy.

tuneslover said,

Because they are free software. So they can carelessly test these kinds of things whenever they want. But Microsoft can't do it. Billion of people depend on it. Windows is not a toy.

Pre-Service Pack 1 Vista users might disagree with you.

Joey S said,

Pre-Service Pack 1 Vista users might disagree with you.

No. Did you get it through Automatic Windows Update? I guess you did not.

Joey S said,

Pre-Service Pack 1 Vista users might disagree with you.

No. Did you get it through Automatic Windows Update? I guess you did not.

Ext3 (and other filesystems used by UNIX/ Linux) are claimed to be fragmentation-free (i.e. they don't need to do defragmentation), I wonder latest version of NTFS (used in Windows 7/ 8) and ReFS are also fragmentation-free or not?

GraphiteCube said,
Ext3 (and other filesystems used by UNIX/ Linux) are claimed to be fragmentation-free (i.e. they don't need to do defragmentation), I wonder latest version of NTFS (used in Windows 7/ 8) and ReFS are also fragmentation-free or not?

Fragment free is as mythical today, as it was when NTFS was delivered and was also billed as fragmentless and not hampered in performance by fragmenting.

NTFS in Windows 7 and Vista are highly fragment-free, but since it is impossible to never fragment data, when alloted space decreases, fragmenting still will occur in the immediate write phase.

NTFS and Windows upper layers that control NTFS are smart, so even if a file has to be fragmented upon a write, if it is heavily fragmented to where it does decrease performance the next time it is read, or any time a file is read and heavily fragmented, Windows 7 will mark it and if able, do a background 'shadow' defrag of the file using the features of NTFS. If this is not possible, due to lack of space and file locations, one a week the OS performs a consolidation and defragmenting process at night. During this phase, it also restructures file location based the physical characteristics of the drive to co-locate sequentially read file patterns and place highly used content in the fastest locations on the media.

As for fragment-free, with linear writing technologies, this is impossible, just as it was when NTFS was billed as virtually fragment free and not hampered by fragmenting.
*NTFS when released did uphold these qualities as much as possible, and with the timeframe of the HDD technologies, fragmentation had to extremely severe to interfere with the performance due tothe indexing and lookup pattern of NTFS. It wasn't until the mid 90s, and HDD technologies were hitting 1gb that fragmentation started to truly impact NTFS in any significant way, which was usually only in servers with far more relocation and read writes than desktop computers.

Desktop users have truly never had 'performance' issues due to NTFS fragmentation except in extreme cases based on the HDD technology in place and the user's HDD being filled for a prolong period of use.

Even in the days of WindowsXP, the only credible cases of fragmentation performance issues were systems that were installed on FAT32 not NTFS, which is more of a problem, as many users never realized and still don't that their OEM installed WindowsXP on a FAT32 partition and sadly not NTFS.

GraphiteCube said,
Ext3 (and other filesystems used by UNIX/ Linux) are claimed to be fragmentation-free (i.e. they don't need to do defragmentation), I wonder latest version of NTFS (used in Windows 7/ 8) and ReFS are also fragmentation-free or not?

The key word here is "claimed". Ext3 can fragment, just like any other file system.
It is possible to limit fragmentation by only placing files when there is enough contiguous free space and adding reserved space at the end of each file, but that's useless for most people.

Heck, NTFS has all of ext2/3/4's features plus many other things such as built-in file compression and encryption...

GraphiteCube said,
Ext3 (and other filesystems used by UNIX/ Linux) are claimed to be fragmentation-free (i.e. they don't need to do defragmentation), I wonder latest version of NTFS (used in Windows 7/ 8) and ReFS are also fragmentation-free or not?

There is no such thing as a fragment free file system - the best you can expect is something that doesn't fragment as quickly due to a variety of technologies (background defragmenting, putting files continuous rather than fragmented. ext3fs/ext4fs both fragment but both are nasty hacks built on top of a previous version rather than something well written from day one.

Aethec said,
Heck, NTFS has all of ext2/3/4's features plus many other things such as built-in file compression and encryption...

Except NTFS has been proven time after time again to be slower (than ext4) _and_ has much heavier fragmentation after extended period of use. You should also compare things like Ext4 disk checking vs NTFS when the structure is damaged and which is easier to recover/repair.

Or shall we take out ZFS and compare that to NTFS? Makes NTFS look pretty laughable in comparison.

daPhoenix said,
Or shall we take out ZFS and compare that to NTFS? Makes NTFS look pretty laughable in comparison.

Please explain how ZFS provides anything remotely useful for the average Joe, assuming Joe has one system drive with only one partition.

daPhoenix said,

Except NTFS has been proven time after time again to be slower (than ext4) _and_ has much heavier fragmentation after extended period of use. You should also compare things like Ext4 disk checking vs NTFS when the structure is damaged and which is easier to recover/repair.

I don't think someone actually cares about benchmarks of a far less capable filesystem that nobody knows vs NTFS which is actually barely noticable in the real world. They could have tested it against the NTFS from 10 years ago too because you can't tell that by name.

thenetavenger said,

Fragment free is as mythical today, as it was when NTFS was delivered and was also billed as fragmentless and not hampered in performance by fragmenting.

While I agree that fragmentationless file systems are a myth, some handle it better than others. NTFS handles fragmentation very poorly, hence the need to include defragmentation tools.
thenetavenger said,

NTFS in Windows 7 and Vista are highly fragment-free

That's incorrect. In terms of the fragmentation to used space ratio, NTFS is probably the worst file system currently in use. You only need to do a fresh install of Windows itself to see that the NTFS file system is immediately highly fragmented.

Windows 7 and Vista even have a service that runs in the background that defragments the online file system. Without which, Windows' performance degrades substantially over time. Compare this to Ext4 for example, and you'll see only minimum fragmentation in relation to the used space. And performance doesn't degrade to any significant degree like it does in Windows/NTFS.

thenetavenger said,

NTFS and Windows upper layers that control NTFS are smart...

If that were true, then it wouldn't need a background defragmenter, now would it?
thenetavenger said,

Desktop users have truly never had 'performance' issues due to NTFS fragmentation except in extreme cases based on the HDD technology in place and the user's HDD being filled for a prolong period of use.

That's simply not true. I've seen numerous cases where NTFS partitions have slowed to a crawl over time. More on XP systems than Vista/7, due in no small part to the background defragmentation service. I've never had these problems with Ext4 or ReiserFS.

Edited by Joey S, Dec 1 2011, 5:36pm :

Joey S said,
That's incorrect. In terms of the fragmentation to used space ratio, NTFS is probably the worst file system currently in use. You only need to do a fresh install of Windows itself to see that the NTFS file system is immediately highly fragmented.

That's the Win32 NTFS driver's fault, not NTFS's.


-The average Joe doesn't care about data integrity ; if somehow one of his holiday pictures has one odd pixel, he won't notice it. ZFS's data integrity features are useful for data centers, not PCs.
-JBOD has no point for your average Joe. Remember, he only has one hard drive...plus maybe one backup drive.
-ZFS can address disks that cannot be created, unlike NTFS which is technically limited by its 64 bit addressing....well...we'll talk about that the day we have exabyte drives.
-ZFS uses checksums and fancy things that take up hard drive space to maintain data integrity...how nice. Joe called, he wants you to explain why his hard drive suddenly lost a tenth of its capacity.
-The rest is mostly technical stuff.

Aethec said,

-The average Joe doesn't care about data integrity ; if somehow one of his holiday pictures has one odd pixel, he won't notice it. ZFS's data integrity features are useful for data centers, not PCs.

I feel I have to object: yes, they do care. A broken pixel in an image is surely a sign of catastrophic failure - a broken pixel is not far off a completely broken file. But, I'm not sure that's much of a differentiating factor to NTFS, because it's not like people actually get those issues much.

Kirkburn said,

I feel I have to object: yes, they do care. A broken pixel in an image is surely a sign of catastrophic failure - a broken pixel is not far off a completely broken file. But, I'm not sure that's much of a differentiating factor to NTFS, because it's not like people actually get those issues much.

If a file is corrupt, ZFS will be able to say "hey, your file is broken".
Then what? Either it's not visible (e.g. broken pixel in file) and the user will not notice it and think the OS is acting weird, or it's visible (e.g. completely corrupt file) and the user will have lost data and blame the OS.

MiukuMac said,

Except NTFS has been proven time after time again to be slower (than ext4) _and_ has much heavier fragmentation after extended period of use. You should also compare things like Ext4 disk checking vs NTFS when the structure is damaged and which is easier to recover/repair.

Or shall we take out ZFS and compare that to NTFS? Makes NTFS look pretty laughable in comparison.

The performance differences are bit over stated.

As for comparing ZFS to NTFS, it is laughable, as ZFS is good theory, horrible implementation.

As for feature to feature comparison, when ZFS was started, it had some gains on NTFS, but with the changes in Vista/Win7, there are truly few advantages to ZFS, for example the 16EiB size advantage that NTFS also supports now.

Addtitionally because of the duality nature of NTFS's model, it supports things ZFS does not and probably will not. These can be important things that deal with things like Compression and other higher layer features of NTFS that do not exist in ZFS.

If you go by the FS comparison guides on most sites like Wiki, it will show that NTFS doesn't support things like last archive timestamps; however, this is a bit misleading as some of these 'features' are irrelevant on NTFS especially when running under NT. Specifically, even look at 'block journaling' which is misleading, as NTFS is essentially doing block journaling, but in a different way. I think the Wiki Site eventually put a * on this for NTFS, but it is still misleading to try to define NTFS by technical defintions of other traditional FS technology terminology. Another example 'Allocate on Flush'... Windows does this on NTFS, but it is not a part of NTFS's specific feature set.

ZFS in theory has promise, but outside of Sun/Oracle's marketing spin, it some super FS beyond the reach of NTFS. Just as the Vista release of NT and the NTFS additions, there is not many concrete reasons Microsoft cannot provide an equal and full complimentary feature set to ZFS - the only limitations are in write structures, which is where the new FS technology Microsoft is working with probably is playing a role in advancing NTFS, as it still behaves and works much like NTFS as a FS model.

thenetavenger

Addtitionally because of the duality nature of NTFS's model, it supports things ZFS does not and probably will not. These can be important things that deal with things like Compression and other higher layer features of NTFS that do not exist in ZFS.

ZFS in theory has promise, but outside of Sun/Oracle's marketing spin, it some super FS beyond the reach of NTFS. Just as the Vista release of NT and the NTFS additions, there is not many concrete reasons Microsoft cannot provide an equal and full complimentary feature set to ZFS - the only limitations are in write structures, which is where the new FS technology Microsoft is working with probably is playing a role in advancing NTFS, as it still behaves and works much like NTFS as a FS model.

Wow. I don't really know where to start here.

ZFS and NTFS just aren't comparable at all, aside from the fact that they are both filesystems.

ZFS is massive in every sense, and amazing in its implementation. NTFS is simple and light in comparison.

For example, you need gigs and gigs of memory to take advantage of ZFS properly. We're running 12GB RAM and a 24TB dataset and could use more RAM really (probably double).

It supports in-line transparent compression and de-duping.
It has checksumming and software raid 6 support (known as raidz2) with hot spares, separate log drives (ZIL) for accelerated writes and read cache drives (L2ARC) for accelerated reads.
It also has realtime, transparent snap-shotting not to mention iSCSI target support, etc.

It's proper enterprise grade storage that can rival real SAN appliances but in actual fact the real beauty of it is in the simple administration. It is a pleasure to use, a world away from hardware RAID and SANs.

To compare NTFS to it is simply ludicrous.

M_Lyons10 said,
Ah... A new name. I guess all the NTFS complainers will be happy...

People complain about the file system itself, not its name.

I think the point is, some people complain about NTFS because it's 'old', not because they know of specific deficiencies.

Neobond said,
Yeah, yet NTFS has been updated several times as well

I believe any people's serious issues with NTFS are mostly in comparison to things like ZFS, which has some features that NTFS doesn't. I'm not all that clued up on it myself, though.

Panda X said,
People complain about the file system itself, not its name.

Most complaints use nonsensical arguments such as "it's 18 years old" or "it doesn't support checksumed checksums even though everyone wants it".

Kushan said,

I believe any people's serious issues with NTFS are mostly in comparison to things like ZFS, which has some features that NTFS doesn't. I'm not all that clued up on it myself, though.

Yep, and ZFS was to be the 'closest' competitor in the history of FS technology and failed to deliver. Even with its success not factored, ZFS had few advantages over NTFS, but lacked several features that have been inherent in NTFS for a long time.

An extensible FS model is far better than a static FS model with all the bells and whistles, which is why NTFS is still relevant 20 years later. NTFS easily adapts and does work agnostically with new technologies like SSD because the extensible upper layers of the FS model allow it to work intelligently with standard object constructs and layers instead of working with generics as most FS technologies, especially when burdened with the UNIX I/O models that we see in Linux and OS X.

Neobond said,
Yeah, yet NTFS has been updated several times as well

But the core architecture hasn't changed since its inception.

Joey S said,

But the core architecture hasn't changed since its inception.

*cough* Linux and OS X are based on a 40-year-old architecture *cough*

Joey S said,

But the core architecture hasn't changed since its inception.

And for you that's necessary? Let me guess, you're an Apple user?

Aethec said,

*cough* Linux and OS X are based on a 40-year-old architecture *cough*

We're talking about file systems here, not monolithic kernel designs, from which I might add, Windows NT is derived.

Windows hasn't significantly changed its NTFS file system for over a decade. FOSS OS's on the other hand, have had many new file systems - Ext3, Ext4, BTRFS, ZFS, ReiserFS, and many others.

Joey S said,

We're talking about file systems here, not monolithic kernel designs, from which I might add, Windows NT is derived.

Wait. You believe NT is derived from UNIX? Seriously?

Aethec said,

Wait. You believe NT is derived from UNIX? Seriously?

Reading comprehension isn't your best skill I take it. "From Which" is the pronoun which refers to the antecedent "A monolithic kernel design", from which Windows NT is derived. Get it?

Joey S said,

Reading comprehension isn't your best skill I take it. "From Which" is the pronoun which refers to the antecedent "A monolithic kernel design", from which Windows NT is derived. Get it?

Still wrong ; NT isn't monolithic.

Joey S said,
Windows hasn't significantly changed its NTFS file system for over a decade. FOSS OS's on the other hand, have had many new file systems - Ext3, Ext4, BTRFS, ZFS, ReiserFS, and many others.

That says nothing about the actual quality of those items.

It can easily be interpreted as: the others keep changing because they're not good enough. NTFS doesn't change, because it is good enough. I'm not saying that's true, just that it's a terrible argument.

Joey S said,

But the core architecture hasn't changed since its inception.

Exactly, because it is an extensible model. Just like NT itself.

There are reason's Cutler's team avoided the traditional UNIX and generic OS models, as they wanted an object based model that could more easily adapt, even though in 1992, this created extra overhead on CPUs of that era, which work in reverse today, as the object interface is many times faster than using smart functionality added on and compensating for a generic model.

NTFS will be viable until the base 'write structures' on the media are no longer able to meet the requirements of the upper level services. Just like NT will be viable for many years to come, as Vista has shown, even a massive shift in the GPU and video model easily slipped into NT, without even having to disable the previous generation technologies.

P1R4T3 said,
There you go, for people saying that NTFS is growing old and that it needs a replacement.

Yeah...now they'll ask a replacement for the NT kernel (which is as old as NTFS).

P1R4T3 said,
There you go, for people saying that NTFS is growing old and that it needs a replacement.

did anybody read the latest article on the Win8 Building Blog? I think they will stick to NTFS

P1R4T3 said,
There you go, for people saying that NTFS is growing old and that it needs a replacement.

Ya, but we don't even know what this is and what it'll be used for. Can it be the default FS fully? Maybe it's like the FS used on WP7 and something for ARM systems mostly? Need more details at this point, though it's pretty interesting either way.

Aethec said,

Yeah...now they'll ask a replacement for the NT kernel (which is as old as NTFS).

Wow really?

Yet it is still newer than OS X's kernel, and far newer than the principles of the Linux kernel...

When you find a better HAL based object oriented kernel technology, be sure to let all of us know how to improve NT, so far the best minds in the world seem to have trouble improving upon it with any significance or in any way that NT cannot adapt. (Think of it like the Borg, by its very nature of design, it is far more extensible and adaptable than is realized.)

thenetavenger said,
Wow really?
Yet it is still newer than OS X's kernel, and far newer than the principles of the Linux kernel...
When you find a better HAL based object oriented kernel technology, be sure to let all of us know how to improve NT, so far the best minds in the world seem to have trouble improving upon it with any significance or in any way that NT cannot adapt. (Think of it like the Borg, by its very nature of design, it is far more extensible and adaptable than is realized.)

You misunderstood me.
I don't think NT is obsolete, and I don't think NTFS is obsolete.
I just think people who go all "let's stop using NTFS because it's old" are idiots, and if they really wanted to be consistent they'd have to stop using NT and other current kernels.

thenetavenger said,
Wow really?

Yet it is still newer than OS X's kernel, and far newer than the principles of the Linux kernel...

When you find a better HAL based object oriented kernel technology, be sure to let all of us know how to improve NT, so far the best minds in the world seem to have trouble improving upon it with any significance or in any way that NT cannot adapt. (Think of it like the Borg, by its very nature of design, it is far more extensible and adaptable than is realized.)

The drive API for Mac OS X and many parts have been been replaced since it moved from Mach 2.5 to Mach 3.0 plus BSD updates but I agree with the overall point you're making. The biggest gripe I have is getting something better than win32 - WinRT sounds nice but it only caters for Metro Applications where as there needs to be a replacement for win32 when it comes to traditional desktop applications.

Mr Nom Nom's said,

The drive API for Mac OS X and many parts have been been replaced since it moved from Mach 2.5 to Mach 3.0 plus BSD updates but I agree with the overall point you're making. The biggest gripe I have is getting something better than win32 - WinRT sounds nice but it only caters for Metro Applications where as there needs to be a replacement for win32 when it comes to traditional desktop applications.

The API shift in OS X is rather insignificant when it still falls back to the base architectural model of operation. Unless MACH is thrown out and something other than the BSD API interface is used, OS X will continue to have limitations that just do not exist on the Windows NT architecture.

Win32 is aged, but it is only a subsystem, that with the MinWin revisions in Vista forward is no longer 'mingled' with lower lever NT layers as had started happening in NT 4.0.

With Win32 it is now more of just a base API set and common subsystem kernel, that is irrelevant even when hosting higher level application interfaces. With WPS/.NET and WinRT Win32 is essentially just a compatibility layer that will slowly disappear for new development.

WinRT is for Metro, but it also is very much a part of the .NET ecosystem, that is the two part portability model Microsoft is making a reality with Windows 8. WinRT may be HTML5 and other technologies, but is still using XAML and is a part of .NET.

Windows 8 takes the original NT vision of portability and using the benefits of NT's lower layer portability and .NET technologies that are architectural agnostic, to create the first comprehensive cross architecture platform for developers where JAVA and other technologies failed because they only targeted either the lower level or upper level of portability.

NT's HAL interface means NT doesn't have to compensate or recode when moving between hardware architectures, where even with Linux's image as 'portable' still requires considerable code changes to conform to the architecture, which often is not completed at an optimal level, losing performance.

NT's method is portable code, but because it is written to a common architecture that the HAL creates for NT, this allows NT's code to highly optimized for the HAL, and the only real work in getting performance is in optimizing the HAL and translation technologies it employs. This is why NT tends to perform better than Linux when ported,, which has not been demonstrated well in the past few years due to the lack of Windows versions outside of the x86-64 and Itanium architectures. Even though by definition the X86-64 architecture is not NT's native architecture, as the HAL architecture does differ.

The better example of why this portability model works well would be WinCE, that using a HAL (OAL) model based on NT, that allows WinCE to easily have an OEM provide the OAL/HAL and run even as a realtime OS, even though it is written to its HAL architecture not the actual hardware. With WinCE, Linux has never been able to compete in terms of performance on devices, and one of the factors of why WP7 enjoys far better performance the Android.

The upper .NET baseline is the second key to why this works well, as it provides a near and sometimes faster than native code performance VM, that is hardware agnostic, again using it is own HAL variation technologies at the application level.

This presents Microsoft with a great opportunity with Windows 8, as they can flip NT to any architecture they want, and with .NET and WinRT based applications, they work without any developer interaction. Which equals true hardware and platform portability at native code performance levels, something that will change computing, and is non-understood or under estimated by Apple and others.

So this is where your dislike of Win32 should be alleviated, as it is truly becoming irrelevant, as developers will not want to stick to the old API set and miss out on all the other Windows8 platforms.

Mr Nom Nom's said,

The drive API for Mac OS X and many parts have been been replaced since it moved from Mach 2.5 to Mach 3.0 plus BSD updates but I agree with the overall point you're making. The biggest gripe I have is getting something better than win32 - WinRT sounds nice but it only caters for Metro Applications where as there needs to be a replacement for win32 when it comes to traditional desktop applications.

Yea.. calm down thenetavenger, it was just a sarcasm. The windows/MS need no defending here.