There’s a dark horse coming up on the outside
Isn’t Sun – and Solaris – almost dead? No and they’re showing quite a bit of life in the storage arena. It is amazing what a $12 billion company can do with a unique strategy and deep engineering smarts.
One big change: after winning the 1.6 billion dollar anti-trust settlement against Microsoft, including a 10 year cooperation agreement, the 2 companies have embraced each other in ways – like storage – unthinkable 10 years ago.
CIFS support in the Solaris kernal
Sun’s steadily falling attach rate led me to give up on Sun as a storage vendor. But the new CEO, Jonathan Schwartz, has a new storage strategy: move storage functionality into the operating system. And there is a VP of Solaris storage software, Bob Porras.
The latest piece: CIFS. For years I’ve listened to engineers moan about the pain of implementing CIFS on non-Windows systems. Now I know why.
In a blog post Sun engineer Alan Wright explains:
There is a common misconception that Windows interoperability is just a case of implementing file transfer using the CIFS protocol. Unfortunately, that doesn’t get you very far. Windows interoperability also requires that a server support various Windows services, typically MSRPC services, and it is very sensitive to the way that those services behave: Windows interoperability requires that a CIFS server convince a Windows client or server that it “is Windows”. This is really only possible if the operating system supports those services at a fundamental level.
Solving those issues required 180,000 lines of new code in Solaris.
It gets better
They also made changes to ZFS (see my ZFS: Threat or Menace?) to support CIFS:
- Support for DOS attributes (archive, hidden, read-only and system)
- Case-insensitive file name operations.
- Support for ubiquitous cross-protocol – NFS and CIFS – file sharing.
Check out the storage community at OpenSolaris to see what else is cooking.
The StorageMojo take
OpenSolaris is becoming the finest storage platform out there. Adding CIFS support to the kernal is a Big Deal: OpenSolaris will be industry’s first OSS universal storage platform.
Only a company with nothing to lose in the traditional big iron storage business could be so bold. My hat is off to Jonathan and Bob.
Update: For more detail on other SMB related changes, check out Doug McCallum’s Share Manager blog.
Comments welcome, as always. Yes, I know the Samba guys aren’t happy. One of these days I’m going to tackle the GPL vs CDDL thing and see if I can make any sense of it. I’m also wondering where all the givebacks are from the storage companies using OSS in their products.
They’ve finally done what NetApp accomplished in ’96? 🙂
This could definitely make the Thumper line more attractive in the long run, but is it going to get to market with the right momentum to make a difference? They need to run a good marketing campaign and have a low price point for entry to get the beach head started. The administration tools need to be easy as well. They could spoof the Geico caveman commercials. “Unix so easy a windows admin could do it!” If it needs any kind of training or highly skilled admins to use, it’ll be DOA. The highly skilled people have already solved these problems by cobbling together other packages.
Robin,
I’m in full agreement — OpenSolaris will be the premiere OSS platform in the near future. There is a lot of interesting projects happening at OpenSolaris.
But, in terms of quality, it still has a way to go. Although OpenSolaris as a storage platform has many innovative features including NAS, OSD and traditional scsi target emulation, many features are relatively untested and many edge conditions are not handled as elegantly as some commercial products. OpenSolaris as a OSS storage product platform cannot be a tool kit, or require significant competency to deploy and operate.
There has been some talk about deploying OpenSolaris and Zfs as replacement for traditional enterprise NAS and scsi raid controllers. OpenSolaris could be “the one”, only problem is COTS hardware still does not have the feature sets to support enterprise quality products for under $1100 per controller. OpenSolaris is a very good first step in making OSS storage products a reality.
I may have mentioned, I’m a big fan of Zfs. I like almost everything about it. Unfortunately, I run windows XP and Vista as desktops. Integration of nfs and logins to unix is not the most seamless in windows. I have one windows server for making my login life easier. Hopefully, OpenSolaris CIFS effort will get rid of this windows box in the near term.
Givebacks from storage vendors for using OSS ? We probably won’t see any from the big 4. Most of their product costs are incurred by poor planning, diminishing domain expertise and middle management talent.
How is someone from lets say the telecom industry going to manage product development and quality when they don’t even know what the product does ? The industry is in real trouble when all managers can do is track defects and occasionally mention what a great job everyone is doing. Its so insincere, its worst than a “dear john” form letter.
Management leadership in the storage industry is becoming eerily similar to the management treads in the music industry. Look at the leadership in 3 of the 4 big record companies: Edgar Bronfman Jr., CEO of Warner Music, previously headed the Seagram Company. Rolf Schmidt-Holz, CEO of Sony BMG since 2006, used to run the German public television station WDR. Until recently, EMI was run by Eric Nicoli, who spent 19 years at United Biscuits. NOTE: “biscuits”
It a tough act to follow, no domain competency, management practices that drive out industry expertise and team talent, the belief employees should work 70 hrs a week for little compensation while executives take ever increasing compensation, and finally rewarding the stakeholders with unprecedented sales and losses and 75% reduction in retail outlets over the last 12 years. Oh, did I mention no strategy to recover the business.
Many storage companies do not value team competency, many believe that one or two experts or a nearly working source code base and a bunch of junior developers in Bangalore will deliver high quality enterprise products to the channel on time. A very poor decision. Do they think they are in the business of building web pages for MySpace users ??
Dell must feel their storage business is at risk using their current storage vendors for them to spend $1.4B on EqualLogic.
Yup, that was another storage industry rant…
On the brighter side, we may see is a new generation of storage vars and products based on OpenSolaris that will supplant the incumbent dinosaurs.
No one is going to really trust open source coders to build storage devices, it’s way too important a business area. It’s more for gamers and hobbyists, not mainstream datacenters. Sun needs to get back to building workstations, and stop experimenting with all these fads.
Rick, yes, when Dave and James started NetApp they had a software business model. That didn’t work too well in the early ’90s, but CIOs who can count are coming around to the idea that commodity HW + the right SW is the way to go. If NTAP isn’t looking hard at that right now – and EMC plans to eat their lunch with Hulk/Maui – they are already behind. Wishing for the good old days won’t make them happen.
Rwhiffen, Sun’s marketing – especially in storage – continues to be a weakness. The fact that they’ve moved storage into Solaris is a mixed blessing. It gets it away from the dysfunctional culture of Sun’s storage hardware group – a big plus. But as Thumper has demonstrated the Solaris folks aren’t ready for prime time either. Overall though I applaud the move as long as Bob hires outside people. After 15 years of a declining attach rate Sun’s storage group is permanently broken.
Xfer_rdy, some early adopter folks are using OpenSolaris in production and bless ’em. I trust the OS community can keep the APIs and UIs coherent as more features are added. Dell knows the SMB storage problem better than any of the other big vendors and the EqualLogic buy is a wise move.
Martin, where have you been? Major data centers are using open source all over the place. Storage is just the next obvious step. Sun still builds workstations? Who knew?
Robin
If I only had a bag o’ gold right now I would spend it on one of two things.
A) Package the Sun offerings into something that can actually be sold. Recruit a hungry and qualified direct sales force – no channel. Build a support group to rival that of NetApp. (The killer support is why I do not mind the price so much.) Put it together with COTS shelves and disks bought from me with support OR if the customer prefers from somewhere else with no support. Are the decision makers at even SMBs ready for that? Is there enough margin to even make it work?
B) Water cooled data centers. Dig a 25 ft. hole in the ground and fill it with flexible pipe much like you see on the Saturday morning home building shows. Run that to a couple of exchangers for what little heat is left after the jaunt through the buried run. Put a couple of room pumps in a support room right next to the datacenter. Then a couple of rack pumps per rack. Do the physical engineering to retrofit IBM/HP/Dell servers, Cisco gear, and the more popular storage. Then meet with those vendors with all the engineering and design work done. Slam dunk. Especially in this age of green datacenters. Direct conduction is WAAAAAYYYYY more efficient than the forced air mess we all have now. With a probable ROI of less than 1 year you could probably sign up most of the Fortune 1000 day one. Someone, other than me unfortunately, is going to retire to their own island on this one.
Robin – spot on, inside Sun we’ve been evaluating upcoming tech trends – see trend #1: http://blogs.sun.com/TA/entry/top_10_storage_technology_trends
> Solving those issues required 180,000 lines of new code in Solaris.
That’s a lot of bugs waiting to be found, unless they just added on Samba.
Kevin, you may overestimate the conductivity of soil: it takes a *lot* of underground pipe just to heat a single house with a heat pump (unless you manage to get down into an aquifer), so dispersing the heat of an entire data center into the ground would probably cost more than using the pipe to take it up to the roof and disperse it into the atmosphere (which may not be all that conductive either but is convective as all get-out).
Unfortunately, your other proposed venture isn’t all that solid either: ZFS is just not the wonder that its developers and CEO would like people to think it is. I won’t repeat why here, but I sketched out some of the reasons in http://storagemojo.com/2007/11/19/nexenta-next-up-on-oss-storage/#comments and in more detail elsewhere (including rather spirited discussions in http://www.opensolaris.org/jive/thread.jspa?threadID=42702&tstart=0, http://www.opensolaris.org/jive/thread.jspa?threadID=44543&tstart=30, and http://www.opensolaris.org/jive/thread.jspa?threadID=45012&tstart=60; there was a fourth related topic, but it had no technical content).
– bill
Bill,
ZFS may not be revolutionary compared to WAFL, but it is included in an general purpose operating system, opposed to NetApp’s systems (you know what app stands for?).
Proprietary systems are going to die, and this is what Netapp is fearing.
Just remember, with (Open)Solaris you can run applications directly on the storage device. Can you do that with a proprietary box?
While ZFS may in some cases be not yet as good as WAFL, it catches up very fast.
Do remember also, a filesystem is not written in two days, and it took Netapp many years to be where they are now. But the gap is closing….
Bill, I think I’ve identified the crux of the disagreement over ZFS.. You value performance over data integrity, while I value data integrity over performance.
I think the market will select data integrity – and ease of management – over performance in most business applications.
Robin
No, Robin: the crux is that you wouldn’t understand a quantitative argument if your life depended on it (fortunately for you, as a marketeer it never has to). As a file system architect my first concern is data integrity – but as a realist I recognize that to customers it only matters up to the point where further improvements yield no *noticeable* benefit, and that’s the realm in which ZFS is operating save for relatively rare installations.
The very fact that NetApp has *not* taken over the entire world proves that most customers are willing to pay only so much for the level of integrity that they and ZFS offer. And costs need not be only monetary: customers won’t pay in terms of performance either unless the added integrity is significant to them.
Another important part of our disagreement is your fanboyish infatuation and complete inability to assimilate information that might diminish it. And if you continue to crop such observations I promise you that I will dog you mercilessly over at your ZDnet blog: you explicitly solicit comments here, so you’d damn well better print them.
– bill
Bill, the essence of marketing deals with human behavior, which so far doesn’t lend itself to quantitative analysis like Ohm’s Law.
Again, where you an I differ is that I believe that ZFS creates a noticeable difference in data integrity and you don’t. Fine.
I further believe that, properly marketed, the ZFS difference can be made palpable to the people who write the checks. You don’t. Fine.
NetApp is still growing faster than the market. Which says to me that people are still discovering their value.
Calling me a fanboy is simply name calling. The people who write the checks decide what succeeds, not engineers. Properly marketed – which is not guaranteed – ZFS will win the check writers.
I edit some of the your more derogatory comments and will continue to do so. If this upsets you then why don’t you start your own blog? You seem to ignore the times where I acknowledge the validity of your objections, even though I don’t hold them to be as important as you do.
Data is cooling. Performance is less important. Data integrity is increasing in importance for reasons I believe in and you may not. Fine. Let’s agree to disagree and move on.
Robin
Bill,
You just don’t get it, do you?
If you can get better data integrity for free (ZFS), why would you pay for worse integrity (WAFL)?
If you can get a better CIFS implementation for free (ZFS) why would you pay for a worse (OnTap)?
If you can run applications directly on your storage device for free (Solaris), why would you pay for one, where you can’t (OnTap)?
1. Believing that ZFS creates a ‘noticeable’ difference in data integrity is not merely a matter of opinion, it’s quantifiable. And the available data suggest that you’re simply wrong: while the difference is certainly statistically measurable (given a sufficiently large test base), it’s in most cases *not* ‘noticeable’ (especially when disk scrubbing is performed, which – e.g. – has been available in Linux for some time).
2. I’ve never in any way suggested that ZFS’s incremental integrity couldn’t be successfully *marketed*: *any* difference, regardless of how inconsequential, can be marketed – your error is in your apparent belief that the ability to market a feature is necessarily proportionally related to its actual utility.
3. If you walk like a fanboy and you talk like a fanboy, either suck it up and accept the appellation or start being more willing to assimilate information that clashes with your preconceptions and to present a more balanced assessment of ZFS. Calling you a fanboy is in fact relatively charitable: from your seemingly endless and uncritical paeans both here and at your ZDNet bloc, it really looks more like you’re angling for a ZFS marketing position (taking you at your word that you’re not currently receiving any form of compensation from Sun).
4. ZFS is by no means the only current contender that can ‘win the check writers’ if properly marketed: one of your more irritating habits is presenting your own unfounded opinion as fact. ZFS has at least one marketing weakness to offset every marketable strength, so in terms of marketing strategy it’s just one marketable contender among several (even if you limit your scope to open source products).
5. Editing to tone down a flame war is one thing; editing to excise material that you’d just rather not see is quite another. If you’re not willing to take criticism for your positions, stop soliciting comments on them; if you think the criticisms should be toned down, then take a more balanced and moderate tone yourself and that will likely happen as a natural consequence. And suggesting that I should start my own blog if I want to comment on what you’ve said here boggles the mind.
6. The fact that data is cooling if anything makes performance *more* important. Data is cooling because so much more is being retained – but the ‘hot’ data is at least as hot as it ever was (and perhaps hotter: the more access people get, the more they seem to want). Without the kind of location management that ZFS so conspicuously lacks, that hot data is dispersed more and more throughout the sea of cool data and performance decreases. Furthermore, every so often someone decides that all that cool data is just begging to be mined, and performance takes an even larger hit.
7. I’ve said it before and I’ll say it again: as soon as *you* move on to other topics, I’ll be happy to as well (surely you’ve noticed that I don’t initiate these discussions: I just respond to them). Conversely, each time you revisit this topic in a manner that I consider unbalanced, if I find it convenient I’ll take the time to present an opposing viewpoint.
– bill
Bill:
What is a ‘fan boy’ and could you ‘quantify’ this? 🙂
I’m not sure why you are so animated about this subject. I believe all sides of the discussion are making good points.
As you know storage demand is booming ~50% per year according to some recent IDC data. I think this in part explains why some of us are so excited about ZFS. ZFS, properly ‘packaged’ and turned into real solutions for real storage use cases appears likely to help enable a move towards commodity hardware, thereby paralleling a shift seen in much of IT over the years.
I personally believe that ZFS is initially best suited for second tier storage (where demand is growing even more quickly than 50%) though it can be used for primary storage if proper care is taken. In second tier integrity is quite important and many aspects of performance less so, along the lines of the point Robin makes above that I think you actually may have agreed with.
If you have other solutions that deal with the ongoing boom in demand for storage, please share them. For me, the option of just buy more from your legacy vendor is a non-starter… but I’d even be open for hearing that as an alternative if that’s your position.
evan