Home | Featured | Enterprise Computing: Netapp The $4Billion Product

Enterprise Computing: Netapp The $4Billion Product

0 Flares Twitter 0 Facebook 0 Google+ 0 StumbleUpon 0 Buffer 0 LinkedIn 0 0 Flares ×

I had a conversation last week with a PR company doing research for Netapp.  This followed just after Netapp released their Q4 results, with revenue exceeding expectations at just over $1 billion.  It’s amazing how in the space of less than 20 years they have developed from nothing to a company selling a single $4 billon product.

Lots of people will be quick to point out to me that Netapp sell lots of products.  Well, yes they do and the majority of those relate to a single core product – Data ONTAP running on some kind of bespoke hardware.  There are a few other bits and pieces out there – DataFort and SANScreen for example, but most software and hardware products still revolve around the core function of providing Networked Attached Storage.

Two thoughts intrigue me:

  • Despite Netapp’s “reputation”, people still continue to buy from them.  By “reputation” I mean, complexity and price – I won’t even mention the sales culture.
  • Competition in the sector must surely mean that growth in the single NAS product can’t continue forever, when newer products that have been developed with the benefit of hindsight are available in the marketplace and those vendors become more established.

It’s the second of these points that probably concerns me most.  Data ONTAP has some technical issues in performance and scalability.  The time taken to develop Data ONTAP 8 has demonstrated that integrating new features into the existing code base is a time consuming and presumably expensive exercise.  Netapp have no other product line to rely on and aren’t introducing new hardware/software as successors to the existing product line.

Compare Netapp to other vendors, specifically their arch-nemesis EMC.  EMC have fundamentally re-invented storage array technology with the introduction of V-Max.  Over the years they invested in technology other than their main Symmetrix range; CLARiiON, Centera, Celerra, Iomega, RecoverPoint are only a few that spring to mind.  There are many more.  The software portfolio of technology unrelated to Symmetrix is even greater.  Netapp remain fixed on their core product platform and the Data ONTAP architecture, attempting to make one hardware device fit all flavours of storage.

Despite the apparent flaws in Netapp’s technology, customers continue to buy and that is reflected in continued growth.  But surely it’s just a matter of time before their market share begins to erode.  Perhaps rather than acquiring technology that further expands features of their current platform (like Data Domain) they should branch out and buy into technology in other areas by acquiring 3Par, Compellent or Pillar perhaps.  Of course the only problem with following this direction is that it admits defeat in using the existing Data ONTAP platform as an all-protocol encompassing storage platform.  When you’ve spend years criticising the competition, you’ve pretty much painted yourself into a corner that becomes very difficult to get out of.

About Chris M Evans

Chris M Evans has worked in the technology industry since 1987, starting as a systems programmer on the IBM mainframe platform, while retaining an interest in storage. After working abroad, he co-founded an Internet-based music distribution company during the .com era, returning to consultancy in the new millennium. In 2009 Chris co-founded Langton Blue Ltd (www.langtonblue.com), a boutique consultancy firm focused on delivering business benefit through efficient technology deployments. Chris writes a popular blog at http://blog.architecting.it, attends many conferences and invitation-only events and can be found providing regular industry contributions through Twitter (@chrismevans) and other social media outlets.
  • http://www.storagerap.com Marc Farley (3PARFarley)

    A company making an acquisition doesn’t have to bend backwards very far to refute their previous competitive claims. “Things have changed!” becomes the new mantra. Besides, sales lives by “don’t look back”.

    Past rhetoric is less important than what the new solution offers customers. Also, a transition can always be handled by letting the newly acquired sales team continue their ways – such as what EMC has done by letting Data Domain rule the roost of their backup business.

  • Anthony

    There is some features of the Netapp Appliance and OnTap software that is not available on other SAN/NAS products. Firstly there is the block level deduplication which can come out with some high deduplication counts. Add this to the thin provisioning and you are already on a winner for virtualisation projects. The NetApp appliances themselves have a storage cache engine (called PAM) which is only now coming to the market via other manufactures like ioFusion.

    If anything NetApp could stop looking at building their own hardware and just buy high end servers from the hardware manufactures and start working harder the onTap software. After talking to some NetApp guys, this could be a possibility. I think they have even been looking at building a virtual appliance also. (Look at DataCore)

    Performance may still be an issue with high IOP but that is not the solution NetApp sell. It is all the other features that their devices can do that is there strength.

    If anything, IBM have now started to rebadge the NetApp appliances and what would stop them from buying NetApp altogether and adding their dedup, thinprovisioning with Cache engine to their DS8000 series storage? (it would be able to provide the high IOPs with the advanced functionality)

  • http://www.brookend.com Chris Evans

    Anthony

    I agree with your comment about de-duplication however thin provisioning is widely available in other products. As for the PAM card, I see that as a workaround for the inability to successfully support SSD.

    It’s certainly an interesting idea for IBM to buy Netapp, however I can see a major culture clash there!

    Chris

  • zax

    20 years of making a product that has features the competition can’t match. 4 billion in revenue from that product? I am not too worried about Netapp. They seem to be addressing most of the concerns you have laid out, and I would venture to say that most customers consider an implementation such as PAM more valuable in the real world than the SSD implementations that are coming to market. Maybe Netapp should buy Xiotech for the ISE product. Now that would be a game-changer…..

  • Pingback: CBTS Launches State-of-the-Art Enterprise Network Operations Center | Network solutions, security and support

  • http://www.brookend.com Chris Evans

    Zax

    I’m sure some of the features (like writeable snapshots) are features the competition can’t match because Netapp hold patents to prevent it (which of course is legitimate business practice).

    So Xiotech….. nice.

    Chris

  • Pingback: How much does an enterprise web filtering solution cost, on a per user basis, for over 5000 users? | Network solutions, security and support

  • http://www.recoverymonkey.org Dimitris Krekoukias

    Ah yes, NetApp, my employer and the company everyone loves to berate, even in the light of us doing better than ever! Interesting how all these posts surface just after the company announces a record quarter and unstoppable year-over-year growth. Right – those are the sure signs of a company in trouble…

    Some NTAP features like deduplication are not available on other platforms – for instance, iSCSI and FC can both be deduplicated and I have a ton of customers on VMware over FC that are getting huge savings. Not so with ANY other vendor. Please don’t belittle accomplishments that nobody else has.

    The large cache is also unique and also dedupe-aware.

    The product is NOT the same as it’s been 20 years ago (the “single NAS product” statement is wrong). This whole article is like saying the current Ford cars are just like the Model T. Or that a company making cars won’t be successful unless they also start making bottle openers. Right…

    Ultimately – the customer satisfaction is amazing, and rarely do people move off of NetApp gear. You can’t say that for most of the other vendors.

    My question to you is – are you truly aware of the capabilities of modern ONTAP?

    Re complexity – have you actually used one?

    D

  • http://www.brookend.com Chris Evans

    Dimitris

    On the first point, yes, I have used – designed and deployed Netapp filers on many occasions. I’ve been involved in discussions where FASXXXX products were pitched at modular arrays for SAN and the FAS product needed twice as much hardware to deliver the same throughput. I’ve been involved in quotes where Netapp undersized an environment to the tune of $1m, expecting the customer to foot the bill for their mistake.

    I am well aware enough on the product to understand the way WAFL uses NVR to understand that this is the achilles heel of Data ONTAP today; I deal with customers who can’t easily understand their capacities and utilisation, even with tools like DFM.

    I see customers who would love to unpick their SnapVault configurations and would clearly move if they could. Technologies like SnapVault, whilst good, tie the customer in and remove any ability to migrate away from the solution until the data naturally expires.

    NAS isn’t like SAN from a migration perspective; the techniques and tools to migrate to other vendors just aren’t there.

    I like your analogy on Ford cars in fact, it may be more appropriate than you think. Whilst the Ford Motor Company did indeed evolve and produce bigger and better vehicles, it was almost bankrupted by continuing to produce large gas guzzling vehicles nobody wanted. Meanwhile the more efficient and adaptable Japanese stole the market from under them.

    Ford is fortunate to have invested in other brands and technologies such as electric vehicles and didn’t need to accept government bailout help to get through the crisis. Perhaps the diversification in their business saved them.

    I’m reminded of two recent events in history which at the time, many friends and colleagues disagreed with me on; (a) the .com bubble, which was obviously going to happen – you can’t build businesses on share price rises alone, they need to make profits and (b) the housing bubble. We all know what happened then. During the good times, there were plenty of people claiming that we would never see a crash that continual growth was here forever….

    Chris

  • http://www.recoverymonkey.org Dimitris Krekoukias

    Interesting how NVRAM is the “achilles heel” of the product, you’ll probably have to explain that one in detail. Maybe you have some suggestions?

    Can’t remember who said it but it goes like this: “don’t just come to me with a problem, also come with a solution you’d like to see implemented”.

    Engineering is always open to suggestions.

    BTW, I have similar undersizing horror stories from other vendors (http://bit.ly/9BHAMF), that’s sales team-specific, not vendor-specific, and IMO those sales people should be fired. I’ve seen it with EMC, Compellent, HP, HDS, pretty much everyone – the reason is human greed and has nothing to do with the products from said vendors (which are usually excellent).

    I’m regularly involved in campaigns where FAS needs 1/2-1/3 (and in one case 1/10th, for a customer with very read-intensive DB’s where PAM worked miracles) of the disks to attain the same performance as the competitors, so obviously our experiences are different.

    I’ll grant you this: If you want pure sequential read performance, FAS will never be more efficient than the competitors. And for some workloads it’s just not the best – nothing is panacea.

    Most tech (especially backup-related) ties the customer in to some extent, the biggest issue I face daily is getting people to migrate, nobody wants to deal with it. So I don’t see any uniqueness there.

    I’ll grant you one more thing: NetApp could be educating customers a lot better about what the options mean and what they do to your capacity. Or, educating them about features (like the SnapManager and FlexClone products) that can transform the way they manage their applications.

    If you only want to use FAS as old-fashioned NAS and/or SAN, you’d better spend your money elsewhere.

    But if you’re prepared to embrace the features on offer, you’re in for a treat.

    Back to capacity: On a box that can do so many things, understanding where capacity goes is just, by nature, not as straightforward as on a box that can only present LUNs.

    You have your WAFL reserve (which can be tweaked to be under 10% but nobody seems to know that), Aggregate snap reserve (which should be no more than 1% but most people don’t touch from the default), then Volume snap reserve (which should be zero for vols with LUNs in them), fractional reserve (should be zero for vols with LUNs in them), your volume and LUN space reservation (if you want thin provisioned you need to set BOTH to “none”)…

    All stuff that doesn’t take me more than 10 min to explain to customers, but I guess not everyone takes the time, and I see a lot of VARs especially not taking the time, focusing on a quick sale.

    Then I get a call from the VAR “the customer is complaining about the available space, please assist”.

    I call the customer, who says “EMC/HP/my border collie was right, where are my space savings?”

    I then have to spend said 10 minutes showing the customer what to do, at the end of which time they are sometimes more upset: “why did nobody tell me this? I wouldn’t have bought all this extra disk!”

    I’ll soon have a new post in my blog about NetApp capacity, since I can prove to you that a box with about 80% usable space is possible without cutting corners. That’s hardly a “gas-guzzler”, especially if you consider RAID-DP protection is factored in. BTW, the 80% does NOT assume dedupe or thin provisioning.

    As with anything, education is key.

    Or maybe the complexity can be explained like this:

    Imagine how much gear a typical NetApp box can replace. We’re talking backup systems, archival systems, replication appliances, extra NAS heads, etc.

    Nobody in their right mind should claim that managing all THAT is actually simpler than managing a single box.

    Unless you have separate people managing each piece, in which case, each piece may well be simpler since you only deal with, say, backups and nothing else.

    But for customers that don’t have enough admins or want to give their admins more time to deal with other stuff, NetApp is a great solution.

    Maybe you should ask Oracle, SAP and Yahoo how many admins it takes them to manage a certain # PB. Then compare to other vendors. Then we can talk about complexity.

    Sorry for the long post, I’m in the bathroom and I always need something to do there… :)

    BTW – see here: http://bit.ly/cgn616 – what’s your bias? Because we all have a bias and an agenda, whether we realize it or not :)

    D

  • http://www.brookend.com Chris Evans

    Dimitris

    You’re right. Everyone has bias. It comes from experience, good or bad. In my experience/bias, I’ve seen plenty of scenarios where Data ONTAP works well as a NAS box. However I’ve experienced more scenarios where Netapp filers have become overly complex than those where everything is running smoothly. That obviously determines my views.

    So, I will take the time to produce a more technical rebuttal on the NVR stuff. When I was first shown WAFL, NVR, RAID-4 etc I could see the simplicity and elegance of it. I could also see the inherent scalability issues too. Expect to see some more in-depth posts in the near future.

    Chris

  • zax

    Random Question;

    At what point do ZFS based solutions become a real challenger to NTAP? I personally see that as one of the largest future challengers to ONTAP. The Solaris team has already released inline and post-processing dedupe in the openSolaris distribution and these features will be rolled into the Solaris kernel soon. Companies are already packaging ZFS (Nexenta anyone?) with easy to use tools and management to create a storage appliance that is hardware agnostic. NTAP will need to evolve. I keep hearing that they are not a disk company, but the sales team’s seem to be focused on selling disk.

  • http://www.recoverymonkey.org Dimitris Krekoukias

    There’s more to designing an array than a filesystem… ZFS has a lot of features (some arguably cooler than WAFL) but an array is the sum of the parts and engineering goes into everything.

    The appeal of OpenSolaris/Nexenta etc? You can load it on whatever, making it appealing for smaller shops.

    Larger shops want the integration, hardened hardware, seamless failover, replication and app integration.

    D

  • SRJ

    Wow…my jaw kept dropping lower and lower as I read through this post – quite a feat for someone other than Barry Burke! =)

    1. NetApp’s reputation?? I am baffled by this… Compared to WHO? You specifically compared them to their “arch-nemesis” EMC. Does ANYONE in the storage industry have a worse reputation than EMC??? I think not.

    Complexity: How is a single well-integrated product more complex than 20 half-baked, non-integrated products?

    Price: Price is relative to value. Obviously customers see value in their solution and think the price is worth it. I have personally seen many customers choose to pay a little more for a NetApp solution versus cheaper competitive solutions because of the additional value they bring. Price != TCO.

    Sales culture: This one is truly unbelievable. I’m sure this varies by geography, but I can’t think of a vendor sales team that I’d rather work with! By contrast, EMC’s sales culture is infamous!

    I think to most readers, it’s clear that your comments on these points are more about your personal experience (quite possibly an exception to the norm), and certainly not reflective of their reputation in the industry.

    Fair enough…

    2. Competition means NetApp growth is doomed? NetApp has seen unbelievable growth for years. They’re still seeing exceptional growth. Has there not been competition in the sector all this time? I’d say that NetApp has been successful at the expense of their competitors…and I see no compelling reason this will change any time soon. Sure, they’re not necessarily the best fit in every situation, but I will say they’re a better fit than most in most situations. Sounds like a recipe for success to me…

    3. Spinnaker code: On one hand you scoff at NetApp for only having a single main product, while on the other hand you scoff at them for taking the necessary time and making the necessary investment to integrate two separate product lines into a single product. Which is it? Do you want two separate, non-integrated product lines, or one? You say they have “no other product to rely on” but why should they need to rely on another product when the one they have is so well integrated and so successful?

    4. EMC “fundamentally re-invented storage array technology” with the V-Max. Uhh…what is fundamentally different about it? You mean how they finally started using commodity Intel hardware after they bashed their competitors for doing so for years? Seriously – what has been “fundamentally re-invented” compared to the DMX-4?

    5. “…customers continue to buy” … “reflected in continued growth”: This is about the only statement in the post I can agree with. But then in the next sentence you theorize that their market share will “surely” erode. Why? I really don’t get it. The market clearly values a well-integrated, high feature/function storage solution and is voting with their dollars.

    Sorry to be so negative…I’m just really baffled by this analysis. I guess the crux of our disagreement is that you don’t see value in the integrated product/solution, but I (and the market, it seems) certainly do.

    You say their market success can’t last. I say they’re just getting warmed up.

  • zax

    A file system does not make a storage product, however Netapp’s biggest differentiator is OnTap. According to most sales guys Netapp is “software company”. It’s going to be increasingly more difficult to charge a high-price for licensing when there are low-cost alternatives.

    Also, I don’t see ZFS based solutions only as only appealing to smaller shops. I work in a fortune 50 company, and we are looking very seriously at different ZFS solutions as a replacement to approximately 1PB of existing NAS storage. We also plan to test ZFS dedupe in our vmware environment.

  • SRJ

    Hold on a sec… While reading through some of your older posts, I just found this:

    “To my knowledge, FAST is the first “innovation” of the new V-Max product line, but it isn’t unique. In fact, I don’t think any features of V-Max are unique; the architecture is found in many other products.”

    What gives?

  • http://www.brookend.com Chris Evans

    SRJ

    Do you believe I’m some kind of EMC fanboi? My comment on EMC fundamentally redesigning *their* technology (i.e.) Symmetrix was to highlight the fact EMC realise that product line couldn’t have continued in the existing form and needed to change to match what customers wanted to pay; Why should I pay a premium for the same disk in a DMX compared to CLARiiON? Disk connectivity is not relevant any longer; shelves have evolved to be reliable and commoditised. Customers no longer want to pay for expensive, bespoke, power eating hardware when mid-range devices are improving and closing the gap. The differentiator becomes software. I didn’t say or mean that EMC had taken the whole industry in a new direction – merely themselves.

    As for my comments on product selection, here’s what I was implying; Netapp purchased Spinnaker purely to integrate into Data ONTAP as the design of ONTAP is fundamentally limited to a loose cluster (of 2) due to the architecture. That was over 7 years ago and we only now are seeing the availability of Data ONTAP 8 “converging” the architectures.

    But wait; I can run the O/S in “7-mode” or “cluster mode”. Does that mean that the code base isn’t really integrated? If I use cluster mode, then I get none of the advanced features of ONTAP? Can I flip between 7-mode and cluster mode? What happens if I start to use the features of 7-mode? Do I have to drop them for the benefit of cluster mode?

    All this seems a workaround. It would have been better to admit defeat, stick with ONTAP 7 and GX and (a) integrate more standard ONTAP features into GX (b) build migration tools to let customers migrate easily from 7 to GX. I’m not aware of any tools existing today.

    Chris

  • SRJ

    Did you not receive my follow-up comment? I’m going to have to stop replying from my iPhone…

  • http://www.brookend.com Chris Evans

    SRJ

    I’ve posted all your comments (good or bad) :-)

    Chris

  • Pingback: Amazon Web Services CTO out to prove enterprise chops | Network solutions, security and support

  • Pingback: Network Security Model – Defining an Enterprise Security Strategy | Network solutions, security and support

  • Pingback: NetApp’s single $4 billion product

  • badkarma

    I wish I had seen this thread earlier. Its front and center to what I’m focusing on right now. I think NetApp is stuck in the mud with their push to 8.x
    In the meantime, the rest of the world is moving forward. Its expensive compared to some of their competitors (EMC for example). There comes a point where one is willing to go with the cheaper and marginally less elegant solution to to save a pile of money.

  • JasonG

    Another ZFS SAN vendor to keep an eye on: Greenbytes

  • http://www.gutgebaut.at/ ivan hallenbau

    a great post, “To my knowledge, FAST is the first “innovation” of the new V-Max product line, but it isn’t unique. In fact, I don’t think any features of V-Max are unique; the architecture is found in many other products.”

    cool, thank you :D

0 Flares Twitter 0 Facebook 0 Google+ 0 StumbleUpon 0 Buffer 0 LinkedIn 0 0 Flares ×