Forget Netapp, is 3PAR vulnerable to "good enough" competition?

In a self-described FUD piece today StorageBod wonders about Netapp's future.  To put a finer point on his thoughts, in a comment he wrote:

Your first billion was the easiest billion; the rest gets harder...there's a target pointed in the middle of your foreheads' now. You are going to have to run even faster, even to stand still. Can you do it and can you do it without increasing your product set?

Normally I would enjoy this sort of badgering, but in a subsequent comment he wrote:

Complaint about NetApp is how vulnerable they are...they are not the only ones tho'...there are more vulnerable companies out there; 3Par, Compellant, Isilon, BlueArc etc, etc...the trailing pack who need to grow quickly or get bought.

OK, at least he mentioned 3PAR first! After all, we are larger than the rest of the pack he mentioned, not to mention in a much stronger financial position. To put a finer point on it, 3PAR is not vulnerable to debt (we have none), nor to cash weaknesses ($108 Million and increasing every year).

So what vulnerability is StorageBod imagining?  I suspect it was similar to those voiced by EMC blogger Chuck Hollis:

But, as you point out, it doesn't strike many of us as a defensible position long-term. It's just too easy for competitors to provide a "good enough" and attack those rich margins.

EMC tended to think it was in an unassailable position a decade ago with the Symmetrix. We were quite wrong. If I was doing a storage startup, it'd be pointed directly at draining NetApp's profit pool.

I also think that single-product companies lose the benefit of hybridization -- combining multiple technologies in new and interesting ways.

My take on this is that the storage industry tends to have different dynamics than most other industries. Yes size matters and Netapp and 3PAR and everybody else plays by those rules, but after achieving a critical mass, size is less important than paying attention to details.  Doing things more effectively  - not just in a more grandiose fashion - makes the largest difference.

We are not concerned about competitors trying to copy us and providing "good enough".  A competitor's assessment of "good enough" will likely not be adequate to many customers. It's how we grew our business over the last 10 years inside the jaws of EMC's formerly unassailable Symmetrix business.  I was an independent outsider when 3PAR began and to tell the truth, I thought the folks at 3PAR were out of their minds. But 3PAR has steadily grown and continues to gain market share - primarily in the high end of the market.  Admittedly, the high end has not been a place for hottest growth over the last decade, but we've climbed and are still steadily climbing.  As the service provider model of IT evolves and gains traction, we like our chances with our products and technologies there. FWIW, our new mid range F-class systems are doing very well, but we probably need a larger reseller channel to address the number of mid-range opportunities we are seeing.  That is a growth problem, not a shrinkage problem.

3PAR's capacity guarantee program works because we consistently deliver utilization levels that our customers didn't think were possible and we stand behind it with a contract.  Customers continue to check it out and are coming to the understanding that great utilization comes from a system designed for optimal efficiency and not simply to be "good enough".  If you stop thinking about storage administration as managing disk layouts you are on the right track. 3PAR admins don't manage disk layouts.

The question of hybridization and crossover potential is very interesting. Innovation can come from trying to tie disparate things together and this sort of integration is incredibly valuable when done well.  However, a hybridization mind set often assumes that most core innovation has already been done and that most future gains will come from integrating across functional boundaries. 3PAR doesn't think fundamental storage technology is even close to being fully developed. The problems customers assume they have to live with should not be tolerated. 

So yes, it's possible that we are vulnerable. The storms of technology change could make 3PAR and Netapp irrelevant in the years to come. Speaking only for 3PAR, we believe there is still a gold mine in revolutionizing storage efficiency and will continue to compete based on that belief. 

Comments

[Stuiesav]

Are 3Par vulnerable - most definitely... Who's is not!!!

At the moment - i think the biggest issue comes from being a FC / Block vendor only...

If i take a look at the way the market is going, the fact that data mobility is king, the fact that 10-gig-e is "out there" and that basically FC is a pain - there needs to be a "what do you do next" attitude.

If i look at my development estate - as soon as 10-gig-e rocks up, and i flood wire my development DC with it, I want NFS... I want portability of data... I want de-dupe... I want not to have to mess around with FC networks - they are painful to manage...

so its obvious what i am asking here... some type of NAS offering..

In terms of competitors - you have HP rocking up with new offerings and a whole heap of others that are building modular scalable arrays - the ability to offer different "personalities" will become more and more common place - it could be a NAS head, it could be an offload engine, it could even be a hypervisor running on a storage node.... - I haven't seen any of these type of scenarios even being discussed by 3Par yet...

Thoughts / comments?

[Stuiesav]

I would also say - that if you dont start doing some of this "other stuff" - lifetime of product and company becomes more and more limited...

[marc farley]

Thanks Stuiesav,

Yes, we certainly have work to do and comments like yours help us sort out our priorities. We don't typically discuss development plans, hallway discussions or knock-down, drag-out fights regarding future products and developments in public.

[Keith Brown]

Regarding your quote from the O'Hollis Factor Marc:

> If I was doing a storage startup, it'd be pointed directly at
> draining NetApp's profit pool.

Dang if that guy isn't insightful! In fact this is already happening. There's a startup out there called Agami, who are "Just like NetApp", only far, far better, and of course cheaper too. They are fast growing, have signed up an enormous number of customers, and you can read all about how great they are at http://www.agami.com, and... oh... hmmm.... that's odd? I guess they must be having problems with their web server today?

Anyway, it's a brilliant business plan, absolutely *brilliant* I tell you. As he clearly recognized this, hopefully Chuck managed to get some money in there? You just can't beat getting in on the ground floor with an opportunity like that eh?

[marc farley]

Keith, the sarcasm might have been a bit heavy. For readers who aren't quite sure what the whole Agami thing is about, it was a NAS company that went out of business rather suddenly a couple years ago. FWIW, I saw one of their systems at Weird Stuff back in November.

There have been many startups over the last 10 years that have had business plans to eat Netapp's lunch and none of them have had much success. My advice for people forming a storage startup today is to design controllers or NAS heads to use with 100% solid state (flash) arrays. In the 3 years or so it takes to get the first technology completed, the cost of flash should be low enough for this sort of system to be viable in the market.

[JohnF]

@Marc

I see you've pushed my predicted SSD viability date out from 12/21/2012 to something in 2013 then.

[marc farley]

JohnF, the viability of SSDs hasn't moved that much, I'm talking about specialized controllers. After SSDs become viable, people will want to do things differently and controller architectures will evolve.

If you are a startup, you have to look where the seams of change are running and guess what the opportunities will be - and cross your fingers that you guessed correctly. Then most of the rest is dumb luck and timing - things that are mostly out of your control.

But re-inventing NAS? There could be a seam there, but probably not with the kind of reward that you need to justify the risks of ruining your life in a bad startup.

[nate]

Kind of OT but it involves NetApp, how bout a blog on this ?

http://www.theregister.co.uk/2010/02/19/netapp_tiering_dying/

My own thoughts have been in agreement with NetApp that I don't think automated storage tiering is the way to go, with the caveat of I have never used such a system nor talked to someone who has(utilizing SSD).. I like the idea of using flash as a really big cache better, of course haven't used this type of system either but at least in my mind it sounds like it'd be more effective. Last I checked the NetApp PAM cards were limited to read operations only so not quite there yet on their gear either.

Or maybe a mixture is good, with automatic storage tiering I worry that by the time the array notices that some data is "hot" and moves it to SSD it may be too late, perhaps 90% of the operation on that data is complete by then.

[marc farley]

Nate, you hit the nail on the head. Slow data migrations like Compellent's aren't going to be very interesting to most customers - and neither are full volume migrations like EMC's FAST does. However, quick, small granularity (sub volume), transparent migrations could be a big win for customers with latency sensitive applications.

An external cache appliance increases the number of physical doohickeys to manage. Maybe that's OK and maybe its a pain. HA comparisons will probably favor tiering as having fewer points of failure. A cache appliance is easier to re-purpose, but is that important? I don't know.

SSD capacities are likely to be sold in smaller more granular increments than cache and would then be potentially more efficient. Tiering does not necessarily require data relocation, whereas caching always does. To me it looks like cache has more elements to go haywire than tiering. The big question for tiering is whether or not the software can fulfill the vision. Time will tell.

None of it matters much if the cost of SSDs don't come down. That seems to be happening gradually, but not as quickly as some had hoped.

[Adriaan]

Marc,

the CH post on "EMC tended to think it was in an unassailable position a decade ago with the Symmetrix. We were quite wrong. If I was doing a storage startup, it'd be pointed directly at draining NetApp's profit pool" was mentioned in previous comments.
I would just like to point out why CH is in Marketing. Notice the masterful way he points out a Symmetrix opportunity and then tells people to go compete with netApp.
I worked for years with sales guys who wanted to go call on a customer who had just bought some shiny new tin from a competitor. They eventually started listening to me telling them to go call on the guys who had not spent their money yet, or on the customers of the vendor your competitor was winning his business from.
So who has the biggest customer base to attack today?
Sorry CH but there are going to be a lot of raids (pun intended) on your comfortable patch, or based on latest news, HP as easier targets

[nate]

I wasn't thinking so much of external appliances but rather integrating the caching into the system. Thinking along the lines of perhaps a triple mirrored RAID 1 set of 2.5" SSDs in each disk enclosure to act as a massive cache. Given SSDs wear out somewhat fast, replace them at staggered intervals, and the triple mirroring of course ensures high availability..

Or perhaps a pair of disk enclosures with nothing but SSD to act as a big cache.

Or something along the lines of FusionIO type technology in the disk controllers, at one point FusionIO was talking about some sort of interconnect between cards which could be used for networking(maybe mirroring over the link)