NetApp would do better to improve the stability of its "high-end" configurations
The war-of-words between Sun Microsystems and NetApp over patent infringement claims wages on, while legal means to settle the matter have failed. The two companies tried to sit down this week for a settlement conference, but judging by Sun's response of publicly boasting that it has removed at least one of NetApp's patents …
Why, would that be the Simon Breden posted all over the Sun blog site? And on the ZFS website too? Hmmm, I detect a subtle hint of bias! No sign of any Simon Breden posting on any Linux sites, but I'm sure he has a wide range of experience.... But let's skip that and have a look at the spec of the box Simon is using for his "home ZFS fileserver":
* Case: Antec P182: $139.99
* Power Supply: Enermax ELT500AWT Liberty 500W SLI $109.99
* Motherboard: Asus M2N-SLI Deluxe $134.99
* CPU: Athlon X2 Dual-Core BE-2350, 45W TDP $94.99
* Memory: Kingston 4GB Unbuffered ECC DDR2 800 (KVR800D2E5K2/2g) $49.99 x 2 = $99.98
* Storage Western Digital 750gb SATA drives, Western Digital WD7500AAKS $139.99 x 3 = $419.97
* Video Card: Nvidia EN6200LE $28.99
* DVD-ROM drive: LITE-ON Black IDE DVD-ROM Drive Model DH-16D2P-04 - OEM $16.99
* OS: Solaris 10 x86 - $0.00
Over $900!!!! I can buy a secondhand REAL server for less than that!!!
Erm..... I think I'll just stick with my old Linux Samba box, which runs just fine on an old Compaq P2! Total hardware cost? Well, the two IDE RAID cards I got for in return for a favour, and the rest was recovered from scrapped systems or skips! Oh, I did pay £8 for the 150W PSU, though. No need for 8GB of RAM or a dual-core Athlon with Samba. In fact, most home fileservers I've seen have either been very cheap recylced PCs or off-the-shelf NAS systems, I can't see many people wanting to spend the amount for the above kitlist just for what looks like a power-hungry system to sit in the background running up the electricity bill. Mr Breden will be getting dark looks from all those green-minded folk if that's what is needed to get ZFS off the ground!
Oh, and Ash, didn't the whole NetApp-Sun love in kick off after Sun went and tried to shaft NetApp with some ludicrous licencing ploy?
IBM N7000/NetApp FAS6000
Simultaneous multiprotocol support for FCP, iSCSI, CIFS, NFS, HTTP, FTP
File- and block-level service in a single system
Upgrade up to 1.176 PB of physical capacity while avoiding service disruption
Support for Fibre Channel and SATA disk drives
16 TB maximum volume size
Broad range of built-in features
Multiple supported backup methods including disk- and host-based backup and tape backup to direct, SAN and GbE attached tape devices
16 GB to 64 GB ECC memory
512 MB to 4 GB nonvolatile memory
Up to 52 full-duplex 10/100/1000 Base-T Ethernet ports
Up to 56 4-Gbps Fibre Channel ports
10-GbE adapter card (optional)
Quad-port 4-Gbps adapters (optional)
Dual FC LVD SCSI Tape adapter
Dual redundant hot-plug integrated cooling fans and auto-ranging power supplies
FlexClone™, FlexCache, MultiStore®, SnapMirror®, SnapRestore®, SnapDrive®, SnapManager® for Microsoft Exchange, SnapManager for Microsoft SQL Server, SnapManager for Oracle®, Single Mailbox Recovery, SnapVault®, SnapMover®, LockVault™, SnapLock® Compliance, SnapLock Enterprise, SyncMirror®, MetroCluster, NearStore® Personality Option, CommandCentral™ Storage, Protection Manager2 , VFM® (Virtual File Manager™) Enterprise Edition, VFM Migration Edition, File Storage Resource Manager, SAN Manager, Operations Manager3, ApplianceWatch™ for HP Openview, ApplianceWatch for Tivoli®
Two of these replicating in real/semi-real time should be enough storage and business continuity for most organisations.
8 GBP for a PSU!? Wild extravagance! Mind you, I out-bloat your solution with my blisteringly average P3 800MHz processor :)
I use a marvellous hot-swappable drive technology (USB), allowing me to add / remove large volumes on demand, as well as exchanging drives when they fail (though they haven't, so far), and a bleeding-edge mirroring technology called rsync. Power consumption is low, reliability has been good so far and performance is ample for 2 or 3 clients at a time downloading patches and multimedia files. Power consumption and noise emissions are wonderfully low as well. The (admittedly manual) fail-over to the backup file-server (which I've never had to do) would take around ten seconds, and I didn't pay for the spare server either.
People with money seem to have no idea what you can do without any in IT these days.
"Oh, and Ash, didn't the whole NetApp-Sun love in kick off after Sun went and tried to shaft NetApp with some ludicrous licencing ploy?"
Yep. Sun managed to buy some patents covering a fairly broad chunk of SAN functionality, and wanted $36 million dollars to license them to NetApp. Plus cross-licensing of all NetApp's patents, which they claimed didn't apply to ZFS (despite a key part of ZFS being nearly identical to NetApp's patented WAFL filesystem) but still wanted the right to use.
Oh, and open-source software doesn't have a license to any of the Sun patents, and could be sued by Sun at any time. As some of them appear to cover key stuff like RAID, anyone wanting to protect open source should be looking for prior art for Sun's patents before they attempt to do it for NetApp's.
Matt Bryant, there you are!
Compare Samba to zfs? What, rofl? ... no really, WTF? Had you come with xfs and lvm, maybe, but samba? I guess you like comparing lemons and apples ... where was the joke icon? I guess you installed HP-UX on that old system.
Mine's the coat with the "HP" name tag (my initials)!
"NetApp would do better to improve the stability of its "high-end" configurations"
Sun should focus on keeping a storage product around long enough to be able to calculate TCO.
SPARC, Solaris - that's it for Sun's stable product offering
Every other offering has aggravated customers by forcing a migration due to product elimination.
Note Java isn't included in the above list because its viability is being questioned.
I don't think NetApp had a chance, PR-wise, being trapped in a no-win situation where they had to choose between suing, being sued, or paying massive licence fees for their own technology.
All Sun had to do was mutter the words "patent troll" whilst dangling ZFS in front of the "Free Software" crowd and all NetApp could rely on from that moment on was a favourable court judgement or settlement in order to prove their case to the masses.
A win for NetApp will harm the credibility of the "Free Software" movement as their morality-based arguments will be undermined by selling their soul to aid a greedy corporation whilst it attempts to crush a competitor.
"....Compare Samba to zfs? What, rofl? ... no really, WTF? Had you come with xfs and lvm, maybe, but samba? I guess you like comparing lemons and apples ..." Well, Simon's system is supposed to be a home fileserver, not a business one (though I have also built a few Samba servers for business uses before), so it's a very valid comparison as Samba is the de facto choice for Linux enthusiasts which Sun want to mug. You may want to check out FreeNAS which can offer Samba for M$ clients, NFS for UNIX and FTP aswell, all in a much smaller and more efficient package than SLowaris Most home fileservers do not need terabytes of data storage, you'll find most home users just want a cheap and simple solution to back up their existing PCs and possibly act as a store for a webserver or media center. Those of us without Simon's (or is that Sun's) budget would baulk at wasting so much money on a home NAS (which is what Simon's system is). And seeing as Samba is a known and trusted solution which runs very well on very little kit, I'd say it's quite a valid comparison to Simon's hulking bit of kit!
"....where was the joke icon? I guess you installed HP-UX on that old system." Jokes on you - anyone with real tech knowledge knows hp-ux is an enterprise UNIX designed for PA-RISC and Itanium, not Pentium 2s! Actually, the old P2 has at various times run Win 3.11, Win 95, NT 3.51 and NT4, Win 2000 (sloooooowly), several flavours of Red Hat and SuSE, White Glove Linux (diskless CD boot with 64MB RAM drive!), a pre-built Linux NAS distro called NASlite that boots from just a floppy, and FreeNAS which boots from USB. I doubt if Slowaris x86 can match their capabilities without a ton of software and hardware bloat. I did download Slowaris 10 x86 at one point, but after seeing the problems it had on modern kit at work I chickened out of trying to find/hack drivers for the old P2 when it was doing so well with Linux!
Curse you and your 800MHz speed machine!!!! Actually, I only paid for the PSU because the Compaq is one of the old ones with proprietary connectors - the swines!
"Yep. Sun managed to buy some patents covering a fairly broad chunk of SAN functionality, and wanted $36 million dollars to license them to NetApp"
Presumably by "managed to buy some patents" you mean acquired StorageTek, who were in prior discussions with NTAP about NTAP licensing some of their patents?
Why ? Because Matt Bryant doesn't want to know about it. Ignorance is bliss.
Why ZFS isn't needed ? Because, to quote from Dave Hitz, Netapp:
"...because ZFS is open-sourced, it lowers the barier to entry for startup
companies to bring products incorporating ZFS technology to market and start competing with NetApp. Indeed, because Sun is distributing ZFS at no cost, it dramatically lowers the product development costs for any company, not just startups."
May be HP's opensourcing of AdvFS will actually help the case for ZFS in finding more prior art on the claims NetApp patents are based upon.
In regard to SAMBA vs. ZFS, anyone with a sane mind would know that they don't compete. ZFS is a filesystem/volume management entity whereas SAMBA is a piece of software for file-serving over SMB/CIFS protocol. There are instances of SAMBA over ZFS in production environment, google on 'samba zfs' has the first hit:
Simon Breden may have used a $900 system, but chances are it would work on most systems as long as the network card and disk controllers are supported. My $200 emachine system with a sempron/1G RAM/320G HD worked fine after I plugged in the network card that came with my DSL kit from AT&T. Plugged in two more 500G drives and 2 case fans to keep them cool, and I have a pretty good home file server for the price. It only needs 80W idle and < 100Watts when loaded. Went ahead and bought a 8G CF card for $35 and a CF/IDE adapter from EBay for $10 incl shipping and installed solaris on it. It has quite a few of NFS & CIFS shares (with SAMBA). Solaris is still bloated by a large margin compared to Linux and has no appliance like packaging, ok but I can easily spare 8G any day to get the full bloat of install and just use the ZFS features for the functionality I get with it. Now, only if I could get a NetApp filer or a Veritas suites, I would dump ZFS in no time.
ZFS is reaching people who don't want to pay for NetApp/Veritas. More add-ons are on the way, here is another ad. for ZFS:
# uname -a
SunOS portal 5.11 snv_74 i86pc i386 i86pc
# prtconf |head -2
System Configuration: Sun Microsystems i86pc
Memory size: 447 Megabytes
# df -kh /
Filesystem size used avail capacity Mounted on
/dev/dsk/c0d0s0 1.6G 592M 1007M 38% /
... so solaris works fine with less than 500MB RAM & an install of less than 600MB.
hint: simply by not installing StarOffice, you save about 800MB :P
So, it's OK for Sun to rip off NetApp because it will make other start-ups have an easier time? As opposed to get Sun out of a licensing hole. Wow, I wonder what you'd say if it was someone "illegally" using Slowaris against Sun's wishes. Someone like Transitive, maybe? Yes, the old Sunshiners certainly only like a one-way street, they're real quick to side with Sun when someone is "taking advantage" of Sun's technology! I'd pretend I'm amazed by your faux outrage but it's getting real boring now.
And Simon's ZFS system is in no way a commercial solution, it's a home fileserver - care for me to spell it out for you? Which means it's "competition" is other home-made fileservers, such as Linux Samba, which seems to run a lot better on a lot less kit. Now, take your head out of Scott's rectum and go read up on Samba. And for anyone considering a ZFS home fileserver try looking at FreeNAS, then tell me one feature you think ZFS has that will make it a better home fileserver solution (Sunshiners wishing to reply - try not to drool too much on the keyboard):
Mind you, whether Sun is ripping NetApp off is still under court trial, and if you go by the latest court updates, NetApp has already lost one parent, and the other five are all under patent reexamination based on prior art submission. We don't yet have complete information whether the claims patented by NetApp would prevail. It might well be that NetApp only did the first commercial implementation of ideas which were already known and ZFS derived from them. In addition NetApp has also been alleged to have ripped Sun off 22 patents, and they have to defend those claims as well. Basically your zealous attitude towards ZFS is quite apparent because no other commercial OS has such a bundled feature which is free, including Linux and your beloved HP-UX. I know you would bring up Linux + volume manager, which is understandable since logic and Matt Bryant don't go together.
Simon's ZFS system is a great example about how a home fileserver based upon ZFS can be built easily. Now, since you know nothing about ZFS or as long as you pretend it's not worthy of your attention, you may well be very happy with buying a $50 NAS box and export them as FAT32. There are plenty of advanced home users who may want to aggregate multiple disks under one zfs storage pool with mirrors, sparing, compression, failover all bundled in, create zfs filesystems in them and export them for use as NFS or CIFS shares. snapshots and incremental backups may not be what you need, but obviously many would appreciate them. I am also sure that you don't know much about FreeNAS as well, because FreeNAS 0.7 has not ZFS integrated, which is quite fantastic - again they are complementary, not in competition.
For other readers, there is a pretty good article here, great read for people who understand:
The current FreeNAS release, 0.69b1, does not use ZFS. There is a development under beta, 0.7, that is using ZFS. If you check a few forums you'll find the most often comment is "Why?". Downloads are massively biased to the non-ZFS version. The vast majority of users see the inclusion of ZFS as unneccessary bloat and are suspicious that a more commercial angle is being pushed, to the point there are rumbles in the BSD community of splitting the tree and going with a non-ZFS version. Thank you for highlighting another example of how unwanted ZFS is.
But to get back to the patents angle - as I understand it, all of Sun's patents it bought with StorageTek are also vulnerable to prior art. In fact, when Sun made the acquisition, both HP and IBM made statements that they were not worried about the patent angles as they were both confidant they had either counter-patents or cases of prior art. Wall Street commentatorts made similar statements when depicting the buy as an expensive mistake, but then I suspect you consider yourself a better analyst than any of those paid professionals.
Sun knew it didn't stand a chance with a patents play against one of the leading vendors, so they decided to try and ambush a smaller player in the hope they would roll over and give Sun a court judgement to try and use against the bigger players. It's just a slightly smarter version of the SCO playbook. NetApp haven't rolled over and Sun haven't got the quick result they wanted and now the added expense of a protracted lawsuit. Sun has given up on innovation and has become what the open source community despises most - a patent troll trying to make a living off other peoples' work.
How unwanted ZFS is ? Wouldn't that be the most plausible recourse when you don't get those features for free in another way. FreeNFS 0.7 with ZFS in still in beta, obviously the most downloaded would be the released version. ZFS is inevitable where needed, biased judgments notwithstanding.
In relation with the Sun/NetApp suite, why is it important for IBM/HP to be in the picture. You can look at the state of the suite, all related NetApp patents are being re-examined, one already rejected; while NetApp hasn't yet challenged any of Sun's 22 counter patents.The fact is, instead of innovating further NetApp chose to engage in lawsuit and obviously expose themselves against Sun's much more extensive defensive patent portfolio. The reality is, Sun is one of the few vendors who have publicly pledged not to use patents against open source, in fact they paid huge sum of money to Kodak until recently for related Java patents - so it's quite obvious who the patent troll is. NetApp thought they would get a quick injunction against ZFS when they started the lawsuit, now they are the one facing a protracted court-battle (Ref: read Dave Hitz's court submission). Sun couldn't care any less. And now with Flash based SSD, ZFS is going to provide majority of features that NVRAM backed WAFL provides, at a tiny fraction of the cost. It's clear why NetApp started all this - they thought the courtroom is the easy way, instead of innovating further and raising the bar.
Sultanas are usually sweeter, and a more apt metaphor for Sun's withering business.
".... when you don't get those features for free in another way ...." What features? I asked for anyone to state what features of ZFS as available in Slowaris x86 made it better than FreeNAS for a home fileserver? In truth, none of the features of ZFS are needed as FreeNAS is already a complete package for a home fileserver. ZFS is just unneccessary bloat, as Simon's hulking heat generator shows. Whilst you might drag up some Sun ZFS marketeering around a commercial SAN or NAS solution, in the home fileserver arena it is completely irrellevant.
"....FreeNFS 0.7 with ZFS in still in beta, obviously the most downloaded would be the released version...." But you say ZFS is just so golly-gosh and mind-blowingly FANTASTIC, so surely all us open sourcers would be just gagging to try it out and downloading the beta? After all, we're all on the mail list for updates. But we're not downloading it. I first downloaded FreeNAS 0.66 when it was a buggy pre-beta, as did many others, because we wanted to see how it worked. There is nowhere near the same interest in 0.7. Just like free Solaris x86 and it's pathetic downloads to support contract conversion rate shows, ZFS and the rest of Slowaris x86 are unwanted by the open source community.
"....instead of innovating further NetApp chose to engage in lawsuit...." Actually, Sun kicked it all off when they went at NetApp with a patent trolling licence ploy backed up by a threat of legal action if they didn't give Sun free access to NetApp's patents. NetApp had a choice - accept the outrageous licence charges and at the same time hand over to Sun their own patents, or go to court. Sun's original plan was for a quick win, but now Scott has to keep hirng more lawyers than coders. And as to NetApp's patents folding, if they were so weak why were Sun so desperate to get access to them in the first place? Because they know ZFS infringes on them.
"....The reality is, Sun is one of the few vendors who have publicly pledged not to use patents against open source...." Rubbish! IBM have made that pledge eons ago, HP have stated they will use patents to protect the open source community, Sun have NEVER made any such statement I have seen, so please provide the source of this amazing missive. Scott McNealy has been open in that Sun's drive to patent trolling is to provide and protect sources of revenue, and he cannot protect Slowaris revenue without using them. Duh! Sun has also made repeated threats against the open source community because they know Linux et al are eating up Sun's business faster than they're old foe M$. Just like SCO and Rambus, Sun are starting with a smaller player to get a judgement to use in future patent wars, and their targets then will be the ones with real money that are currently helping Linux stuff Sun down the garbage shute - HP, IBM and Dell. After they get licencing deals out of them, Sun will be free to shaft the Linux disties. It's all about competiting by litigation because Sun can't keep up in the innovation game by any other means. Face it, Sun has become the new SCO.
"....And now with Flash based SSD, ZFS is going to provide majority of features that NVRAM backed WAFL provides, at a tiny fraction of the cost....." Last time I checked, SSDs weren't exactly cheap. In fact they're available right now - we have a pair of Texa Ram-San 500s we're using alongside EVA8100s for an Oracle project. No ZFS in sight, no Sun kit in sight. Yet another innovation that has nothing to do with Sun or Slowaris. In fact, we've had shoot-outs and benchmarking sessions to show that the EVAs were a better choice than the NetApp kit, again without any ZFS or Sun anywhere in the picture. Sun was then and is now completely irrellevant, much as your statement is. There is nothing to stop NetApp using SSD technology. NetApp's challlenge has been moving from commercial NAS to SAN, so ZFS is the least of their worries - they have a harder time competing against the established SAN leaders EMC and HP. Sun has zero chance of innovating anything in the SAN arena, all their SAN kit is badged, so they've gone off patent trolling instead.
So, I expect all the Sunshiners will have to start living under bridges and picking on passing billygoats, and just hope that NetApp isn't a bigger Gruff than they thought!
Try again, ZFS does not need much more resources than a standard x86 box, it just needs tiny bit more memory, so your hypothesis that you can't run zfs in ANY x86 box is blatant lie - it's called distortion of facts. But why am I complaining, because nothing else is expected from someone who never tried it and yet claims to possess all knowledge about it. But again, ignorance is bliss.
No, Sun didn't kick off the litigation, they were only continuing the patent discussions that were already ongoing between Storagetek and Sun. The fact is that NetApp had approached NetApp with interest in purchasing some of Storagetek's patents, and storagetek wan't ready to sell them. Once NetApp saw Sun buying storagetek, they saw this as an opportunity to not pay for the patents that they were in discussion with the old company and instead go to courtroom.
"Sun have NEVER made any such statement " - again since you are supremely ignorant, only GOD can help. But hey, why don't you quote your sources when claiming some of your blatant and distorted proclamations!
Sun has never filed a lawsuit against anything open source irrespective of license - that is the fact today, so keep your speculation and ramblings to yourself until you can prove (not state) otherwise.
No, NetApp should not be worried about Sun and ZFS, they need to be worried about the smaller vendors building ultra-cheap AND RELIABLE storage boxes, and SSD provides that opening where these boxes can have majority of the advantages WAFL provides. Yes, SSD isn't cheap, but ZFS's planned usage of SSD (or rather flash storage) is not as primary storage, but as an additional level of cache, very similar to how NVRAM is used by NetApp. So it's still very very cheap and lifts performance an order of magnitude. Not every application needs the expensive storage arrays from NetApp, EMC where you have to pay huge premiums for every additional TB you use - but hey !! these are not my words, that's Dave Hitz declaring in court - or you can claim he may be just joking...
But you are quite happy in your own little world of Itanic,HP-UX,EVA.... be happy. There is a big world of new Web 2.0 crowd unfolding before your eyes, and these companies don't need EMC, NetApp for their core business, cheap reliable storage is all they need. That is were ZFS shines, but you wouldn't pretend to understand that anyway...
So yeah, you keep hoping Sun would disappear one day, HOPE is what keeps us afloat.
"....ZFS does not need much more resources than a standard x86 box, it just needs tiny bit more memory...." There is no such thing as a "standard x86 box", x86 is the hardware architecture. And if ZFS needs more RAM how can it not need more resources - either it does or it doesn't. From what I have seen in the real business use, it needs much more resource and produces poorer performance compared to Linux. But then I don't really expect you to have any real world experience.
"....nothing else is expected from someone who never tried it...." No, I haven't tried it at home because I have no need for it, Linux and BSD-based solutions already offer a tried and tested solution without the bloat. And at work it's the resident Sunshiners whom have lumbered themselves with trying to produce a Slowaris 10 and ZFS alternative to our production Red Hat systems, so far with zero success. It is their experience with poor hardware compatibility, poor application availability, nightmare porting issues and poor performance compared to RHEL - if your own supporters can't make it work why do you think I should bother when Linux already does the job.
"....Sun didn't kick off the litigation...." I didn't say they did, I said they went patent trolling. Try looking the expression up. Ask an adult to explain the longer words for you.
"....The fact is that NetApp had approached NetApp...." What? NetApp were trying to buy patents from themselves? NetApp are suing themselves? You are obviously a very confused individual.
"..."Sun have NEVER made any such statement "...." Nice bit of selective cut and paste, I actually said "Sun have NEVER made any such statement I have seen". And if you're expecting God's help, as an agnostic I'd have to say you seem to be suffering from delusions of two cults, one being that of the Sunshiners and the second probably of Scientology. But, I note both your quoted articles relate to the "open-spurcing" of Solaris x86 back in 2005, when Sun dumped its unwanted code on the open source community. All the patents refer to Solaris x86 code and had to be released to open-source Slowaris x86, so not a contribution to anything other than trying to desperately generate a Slowaris x86 community to rival Linux. Neither article has a statement saying Sun promises not to use its patents against Linux, in fact the second article makes the specific point they HAVE NOT made any such pledge. In short, you have simply proven my point. Debating with you is beyond shooting fish in a barrel, it's more like I put the loaded gun in the barrell and your piscean self plays Russian Roulette!
"...NetApp should not be worried about Sun and ZFS...." Really? I really hope you do work for Sun, hopefully in a senior management role, as you clearly are completely clueless. So, NetApp see a company copying the core code from their product, then open-sourcing it, and then that same company tries to rip them off with a dodgy patent mugging, and you suggest they just ignore it? As I said in my previous post, NetApp's commercial worries are that it is trying to grow out of the NAS market, where it has been very successful, into the SAN array market, where it is finding it much tougher. And as for "ZFS's planned usage of SSD", using high-speed memory as cache in front of disk arrays has been around for years, long before ZFS was even created. Search the Reg and you'll see some old posts on solid state disks. You have to realise just because Scott says he created everything first or he was the first to think of something, doesn't mean it's the truth.
"....Sun has never filed a lawsuit against anything open source irrespective of license - that is the fact today, so keep your speculation and ramblings to yourself until you can prove (not state) otherwise....." Oh, ye of short memory! Did you hear of a certain deal between M$ and Sun back in 2004? The one where Sun got immunity from M$ for Sun's StarOffice users, but happily signed up to allow M$ to sue open source users of the Open Office code in future? Sun knew what they were doing, they hoped M$ would do the dirty work for them and leave them as the only alternative to M$ Office using the code from the Open Office project. Luckily, M$ haven't used it yet, but if they ever do then the blame lies with Sun.
But let's hear from arguably the most important guy in the whole Linux community what he thinks of Sun. Linus Torvalds put it like this in June 2007:
"Don't make the fundamental mistake of thinking that Sun is in this to actually further some open-source agenda. Here's a cynical prediction but backed up by past behaviour of Sun. First off: they may be talking a lot more than they are or ever will be doing. How many announcements about Sun and Linux have you seen over the years? And how much of that has actually happened? They may like open source, but Linux has hurt them in the marketplace. A lot. Not only drools after our drivers, they drool after all the _people_ who write drivers. They'd love to get kernel developers from Linux, they see that we have a huge amount of really talented people. So they want to talk things up, and the more "open source" they can position themselves, the better. They may release the uninteresting parts under some fine license. See the OpenSolaris stuff - instead of being blinded by the code they did release under an open source license, ask yourself what they did *not* end up releasing. Ask yourself why the open source parts are not ready to bootstrap a competitive system, or why they are released under licenses that Sun can make sure they control. So the last thing they want to do is to release the interesting stuff under GPLv2. Quite frankly, I think the only really interesting thing they have is ZFS, and even there, I suspect we'd be better off talking to NetApp, and seeing if they are interested in releasing WAFL for Linux."
Can you feel the love? I especially like the bit about WAFL vs ZFS, completely shows up the drivel sprouted by Sunshiners that Linux likes ZFS!
I am happy in my (actually quite large commercial enterprise) world of Itanium, hp-ux, EVA, WIndoze and Red Hat (oh, and XP arrays, and hp blades, and don't forget the hp printers - gotta love those MFPs!). How's your Mcjob going?
ZFS doesn't use the extra resources for nothing, because it provides end-to-end data integrity, so it can provide the reliability similar to a EMC/NetApp box using all standard components at a fraction of the cost. Since you don't understand ZFS at all, let alone how file-systems work, you should have already known that performance on a single disk is not ZFS's focal point, the bigger the number of disks, the scaling and reliability of ZFS shines.
I meant NetApp approached StorageTek to buy StorageTek patents, but you very well knew what I meant. So try another route. The facts are clear, when NetApp couldn't get access to StorageTek patents, they sued Sun in the pretext of ZFS, hoping Sun would simply hand over the StorageTek patents for free. Sun as a company never initiated any patent conversation with Sun.
"Sun have NEVER made any such statement I have seen" - Nice try. That's the reason I said ignorance reins supreme in you. If you didn't see something, how are you confident enough to proclaim your lies !!
I think you are a real anti-Sun fanatic. Did you actually see what those 1600 patents are - before claiming they are related to solaris x86 code ? No, most of them are related to fundamentals in computing. And Sun did make the statement that they made them available to any open source code. At least these patents are way more valuable than IBM's patent publicity stunts that even included patents related to screwdrivers!! Again, the fact is that Sun has never sued any open source product, NetApp did. So keep your conspiracy theories to yourself until you show evidence as of today.
"but happily signed up to allow M$ to sue open source users of the Open Office code in future?" - Oh... sure you seem to have been the third party verifier for that agreement. Care to provide reference, keep invent lies after lies.
"using high-speed memory as cache in front of disk arrays has been around for years, long before ZFS was even created" - I think someone should really ask you not to try to crack something that's beyond your area of expertise. A flash is not a high speed memory like RAM, and secondly it's persistent memory, unlike RAM. There is a different functionality and economics involved because you can add 32GB of flash cache at almost negligible relative cost addition and ZFS can speed up the read/write transactions by an order of magnitude. It's not something Sun had to invent, the fact is that the architecture in ZFS allows this speedup with no changes in it's architecture. BTW, can you care to do some research and provide reference here who is trying to use SSD as disk cache, try hiding behind those kind of pretexts. It's one thing putting SSD in place of disks and another thing to have an architecture in place that can use SSD where it really shines.
"NetApp see a company copying the core code from their product, then open-sourcing it, and then that same company tries to rip them off with a dodgy patent mugging, and you suggest they just ignore it?" - as I said earlier, the real reason NetApp sued is to get free access to StorageTek patents. And what core code ? The court says otherwise. The patent office is not sure the so called 'core code' has plenty of prior art, and now those patents are on their way to oblivion.
Ohhh, so you did seem to be finding time to actually look for what Linux is saying, and oh my !! He seems to find ZFS interesting enough for Linux - doesn't that contradict what you have been saying so far. I am sure he made that statement before the Sun/NetApp lawsuit, because if he knew, he wouldn't have made that statement. A company like NetApp would only open source code relevant to their products, not their proprietary WAFL code, because Dave Hitz has now declared in court that doing so would jeopardize their existence because it would allow smaller vendors to offer the same features at much cheaper entry point. So now you feel the frustration Linus has ? He can't get ZFS, and WAFL is a far far cry.
Your quite large commercial enterprise must be happy with you for saving them money I am sure - you must be a religious follower of the pledge - nobody gets fired for buying IBM. The companies I mostly work are not so rich they don't even have money to buy NetApp or EMC or Hitachi.
Where to start, there's so much humourous material in your last post!
"ZFS doesn't use the extra resources for nothing..." I'm glad to hear it, Simon's home solution sounds pricey enough, I'd hate to think how much you'd have to spend to get a viable commercial solution.
"...it can provide the reliability similar to a EMC/NetApp box...." Ha! So ZFS all by itself provides all the hardware redunancy, mangement and provisioning features of an EMC array like a Symmetrix? Yeah, right! Sun wish they could make a piece of storage kit with that level of capability, then they wouldn't have to resort to badging everyone elses kit. You really don't have a clue as to what you're talking about.
"....Sun as a company never initiated any patent conversation with Sun...." You're still obviously very confused! Try taking a few deep breaths before next assaulting your keyboard. Actually, take a whole lot of long breaths, the rest of us won't miss you.
"...."Sun have NEVER made any such statement I have seen" - Nice try. ...how are you confident enough to proclaim your lies !!...." Nope, not a lie, just a statement of fact. I haven't seen such a quote, you were unable to provide such a quote (in fact, one of your two examples disproved your own statement), so how am I lying? It seems that whenever your blind faith cannot be backed up by any proof you resort to childish name-calling. All very entertaining but unlikley to win you any converts to the Cult Of The Sunshiners.
"....I think you are a real anti-Sun fanatic...." Yes, through bitter experience, I am anti-Sun. But you seem to have all the traits of a quasi-religeous fanatic - unquestioning belief despite obvious and continual proof to the fallacy of your beliefs. Do you think The Great God Scott is going to make you Saint Fatso if you bleat long and loud enough? How quaint.
"....Care to provide reference, keep invent lies after lies...." Try a web search for Microsoft, Sun, Open Office, or just look at:
"....you can add 32GB of flash cache at almost negligible relative cost addition and ZFS can speed up the read/write transactions by an order of magnitude..." I'm sure it can, but then using flash will speed up all disk access through the controller, regardless of whether it is using ZFS, ReiserFS, LVM or any other filesystem. This is not a feature of ZFS, it's a feature of the use of flash technology. So adding it "at negligable cost" (what, no development costs?) is a common for all filesystems. Do you want to claim that switching from single-phase power to three-phase gives savings due to ZFS too? I really hope you're the Sun bod that is drawing up the new Sun feature sale sheets, at least they'll be humourous!
"....The court says otherwise..." Actually, the court has said nothing of the type yet. Oh, sorry, I forgot, you are accepting the judgement of The Great God Scott as gospel and a higher law than the common law of the land. It will probably be a while before examination of the ZFS code and the WAFL code and associated patents are completed, probably years going by the pace of the similar SCO case. But, Sun's own engineers have admitted they designed a WAFL-like solution after looking at NetApp's WAFL, so they've kind of shot themselves in the foot to start with.
"....Ohhh, so you did seem to be finding time to actually look for what Linux is saying,...." Yes, I did. You obviously did not. Let me summarise what Linus said - there is nothing in Slowaris we need, the only bit that might be of interest is ZFS, and he'd prefer to have WAFL open-sourced than go with the already "open-sourced" ZFS. Oh, and he also said that Sun were just playing at open source to try and improve their market position, and not to trust or believe what Sun say. Care to disagree with that summary? Am I surprised NetApp don't want to open-source WAFL - not really, there is no reason for any company to release proprietary code unless they want to. You may remember that Sun hadn't released Slowaris as "open source" (faux Sun style, not GPL) until their business was going down the toilet. It was never even mentioned as a possibility when Sun was riding the dotcom wave. Pot, meet Mr Kettle. Advice to Mr Kettle - if you shake hands with Pot, count your fingers afterwards just to make sure they're all still there!
"....you must be a religious follower of the pledge - nobody gets fired for buying IBM...." Actually, whilst we have a range of vendors' kit, we are predominantly an HP shop. The big clue (to anyone with technical knowledge) would have been that of all the kit I listed, none were IBM products....
Biting the hand that feeds IT © 1998–2019