NetApp - Dave's Blog

Mark Lewis's Blog as an Excuse for Blog-Introspection

I'd like to welcome Mark Lewis to the blogosphere. He is the "Chief Development Officer" at EMC, so we've now got executive blogs from all three of the of the big "storage focused" vendors: Mark Lewis's for EMC, Hu Yoshida's for Hitachi, and of course mine for NetApp.

I started my blog just over a year ago, and so far I've talked little about the blog itself. I figure an anniversary is an opportunity for introspection, and Mark's blog is an extra excuse.

I can say that blogging has turned out to be more work than I expected. I've written a lot in my life—and enjoy it—but never before on a weekly schedule. I've mostly met my goal of writing an entry per week, usually six or seven hundred words, so after a year I've written over 30,000 words, a good fraction of a book. (A typical book is 50,000-100,000 thousand words.)

In his first entry, Mark made an interesting comment: "Is this a corporate Blog? NO." In some ways, his comment makes no sense. I mean, his blog is hosted at emc.com, and he's an EMC executive talking about EMC products, EMC strategy and the storage industry. Not a corporate blog?! Even so, I think I understand what Mark is trying to say, because I am aware of the distinction in my own blog between Dave's position and NetApp's position.

Even though I am a NetApp executive, what I write is not NetApp's "official position". The main reason is that it doesn't go through the normal checks and filters that we use for press releases and other official statements. I obviously don't want to get into a public fight with my own company, so it's not like I'm going to go crazy. Even so, this lack of formal process gave the legal folks serious heartburn when we first started.

The compromise is that the PR group always reads my blog before it goes up, just to double-check that I don't preannounce anything, or violate SEC rules by saying the wrong thing during a quiet period. If I post about a particular product or group, I usually run it by the appropriate folks before I put it up. But I don't necessarily take their advice!

One difference between my blog and NetApp's "official position" is that I am willing to acknowledge the strengths of competitors in ways that make most marketing folks uncomfortable. There are many strong competitors in the storage industry, and I don't think it hurts NetApp's competitive position for me to acknowledge that fact. I believe it would hurt my credibility to pretend that we are the only company with good products.

I call this my "Miracle on 34th Street" strategy. For people who haven't seen that movie, a 1947 classic, the basic plot is that the real Santa Claus gets a job at Macy's playing the role of Santa Claus, posing for pictures with kids and asking them what they want for Christmas. He almost gets fired because he keeps sending parents to other stores to buy the hard-to-find toys that their kids ask him for, but just in time the President of Macy's comes by and says, "Great job. What an idea, sending parents to other stores! The parents love it and its great publicity."

My favorite blogs are the ones where I can give an insider's view of how NetApp executive management thinks about strategy. What were we thinking when we acquired Decru? Why did we sell NetCache? What were our goals with the StoreVault division? A related set of favorites talks about how industry trends influence strategy. Do we worry about flash memory replacing disk drives? Why do ATA drives require a double-failure protecting RAID? What is the role of iSCSI, and is it disruptive?

Finally, I've enjoyed comparing and contrasting EMC's strategy with NetApp's strategy. Hopefully Mark Lewis's blog will give me more material to work with!

Buyers Are Liars

The customer is always right, of course, but doing what customers want can still be challenging, because they lie. I first heard the phrase "buyers are liars" from a frustrated salesman. Many salesmen have encountered customers who say, "If only you would do X, then I would definitely buy," but then when X is done the customer still does not buy. Product developers sometimes get involved when X is a request for a particular feature.

Our CTO, Steve Kleiman, http://investors.netapp.com/biodetail.cfm?BioID=1586 gave me a great non-technical example of this. He was house hunting, and he told his real estate agent that he wanted a house with no lawn or plants because he hates gardening. She kept showing him houses with brick and cement yards, but he found them cold and sterile. Finally she took him to a house that completely violated his request—plants and grass everywhere. When he complained, she said, "Don't worry, I found you a gardener." Sure enough, that's the house Steve bought.

In this case, Steve was the liar, and his agent made the sale by listening carefully to his concern, but not listening so closely to his proposed solution. If she had believed him, she would probably never have sold him anything.

I often think of this lesson when I hear NetApp customers describe a feature they wish we would add. Even if we implement exactly what they say they want, they might not be happy. The problem is that customers are experts in what bothers them and experts in understanding problems in their environment that they need solved, but they aren't necessarily experts in product development and they certainly aren't experts in the implementation details of our products. Even good developers who know our technology well will often require a few tries to define just the right feature to solve a particular problem. Why would we expect customers to do better?

How do you deal with this contradiction when on the one hand, "the customer is always right," while on the other hand, "buyers are liars?" My compromise is to assume that they are right about their own problems and frustrations, but I don't assume that they've designed the best possible solution.

When a customer requests a feature, I think it's important to dig deeper and ask lots of questions. What problem are you hoping to solve? Exactly how were you planning to use this feature? Are there other features that you considered that might also help solve the problem? If done well, I believe that this kind of interaction can result in solutions that work much better than they would if either customers or engineers specified the feature on their own.

I Admit that NetApp Isn't the "Top Ranked" Network Storage Vendor

In a recent blog, I excitedly pointed out that NetApp had just passed EMC in terabytes of "networked storage" shipped.

That is true, but it also raised several questions. One blogger pointed out that Dell sells EMC storage, and that EMC's number would be bigger if you included that. Another reader pointed out a press release that said, "HP Climbs to No. 1 Position in Total External Disk Storage Systems", and asked the difference between that claim and mine.

Whenever you look at market share, some questions to consider are:

1. What market should you look at? (Networked Storage or all External Storage Systems?)
2. Which vendor gets credit when there is a partnership? (EMC or Dell?)
3. What should you measure? (Terabytes or Dollars?)

Non-technical examples make things more intuitive. To compare car companies, should you count only sedans, or should you include minivans and 18-wheelers? Usually it's best to define a market in ways that capture "natural customer behavior". What alternatives would a customer normally consider together? You would expect a parent looking for a family car to consider both sedans and minivans, but not 18-wheelers or bicycles. Should you count dollars or cars? Rolls-Royce collects lots of dollars, but they don't sell many cars. It's not that a particular answer is right or wrong, but you learn different things by looking at data in different ways.

I defined Networked Storage to exclude mainframes and DAS (direct attached storage) because mainframe storage and DAS feel different from SAN, NAS and iSCSI in the same way an 18-wheeler feels different from a car. People buying mainframe storage wouldn't even contemplate iSCSI or NAS, but people buying SAN often do. Why does HP rank so high for storage overall? In the past year they sold over $3 billion in DAS—over $2 billion in systems that cost less than $5k. See what I mean about DAS feeling different than Networked Storage? This is mostly disks bundled with windows servers. Barely any networked storage even sells for less than $5k.

Another tricky question is how to handle partnerships. When Dell sells a CX300, should Dell get credit or should EMC? IDC always reports numbers according to who sold the system, so they give Dell the credit. This makes sense if you consider markets from the perspective of sales channels. Dell "owns" the customer, so Dell gets the credit. On the other hand, technology people are often more interested in the technology, so they would rather give EMC the credit. I went through IDC's data product by product to figure out who actually built the technology. Lots of companies sell other companies' storage, but the big ones are Dell selling EMC, HP selling Hitachi, and IBM selling Engenio, all over half a billion in the past year.

To help understand the storage market, I built the chart below. It ranks storage companies according to who sells the storage (Vendors) and also according to who provides the technology (Tech Suppliers). It measures vendors and tech suppliers both by revenue, because dollars are always important, but also by terabytes, because that indicates how much customer data is actually being stored. Finally, I added a "blended" score that gives equal weight to each of the other four metrics. (All of this is based on IDC's 2006Q2 data, and includes SAN, NAS and iSCSI, but excludes DAS and mainframe.)

You can see that EMC (dark blue) is #1 for every metric except "vendor terabytes", where NetApp (yellow) is very slightly higher. EMC wins by the most in "technology supplier revenue", because that includes EMC storage sold by other companies like Dell.

These rankings tell an interesting story about companies and their strategies. IBM, Dell and Sun all drop dramatically from Vendor to Tech Supplier because they sell so much of other people's storage. EMC and Hitachi go up. Engenio sells only through partners, so they don't show up at all as a vendor. EMC's position in the market allows them to charge a premium price, so they score especially high when you measure by dollars, but lower when you look at how many terabytes the customer gets.

The blended metric puts NetApp in third place, pretty close to HP, but both of the dollar-based rankings put us in fourth place, so I certainly wouldn't claim that NetApp is #1 overall! What got me excited about the new IDC numbers was that NetApp had passed EMC by any broad market share measure.

Here's the message that our terabyte ranking should send to EMC and HP:

Warning: Objects in mirror are closer than they appear.

[All numbers are from IDC's "Worldwide Quarterly Disk Storage Systems Tracker, Q2 2006 Release", August 2006. The "tech supplier" numbers are based on my own categorization of product lines. IDC does not break data out this way.]

Ten Years of Vision, From IPO to Fortune 1000

NetApp went public a bit over 10 years ago, so that makes it about 10 years from IPO to Fortune 1000.

I just stumbled across an internal e-mail that I sent to all of NetApp in 1996, shortly after the IPO. I thought you might be interested to see what our goals were way back then. It's amusing to see that Sun still had 35% share of the NFS market. We were just starting to branch out from UNIX and also target NAS for Windows. SAN and iSCSI were so far in the future that I didn't even mention them. Still, I think you can see how these goals laid the foundation for what we have become.

When Tom Mendoza, president of NetApp, gives talks to our new employees, one of the things he tells them is that they should write down their goals. Something about putting your aspirations in writing makes them more likely to come true. I think the same is true for company aspirations as well.

Here is the e-mail:

From: Hitz, Dave
Sent: Wednesday, February 21, 1996
Subject: What's next for James and Dave

We've had several people ask "what's next for you," both inside the company and outside. I've had VCs and head-hunters call with the "next great startup". In some ways, this may make sense. Many people do view an IPO as an ending point. During most of our time at Network Appliance, the IPO was a distant target that we mostly ignored as we aimed at nearer goals, like fixing the next bug or shipping the next product.

So having reached the IPO, maybe it does make sense to ask, "What next?" And yet, when James and I think back to what we thought we could do when we first started Network Appliance, we realize that we've barely begun to accomplish our original goals:

(1) Convince the industry that network file service should be handled by filers, and that NetApp filers are the ones to use.

We believed that filers and the appliance philosophy could change the way people view network file service in the same way that Cisco routers changed the way people viewed network routing.

For people to understand the appliance philosophy, they must hear about it. For people to choose NetApp, we must design and ship the best filers possible, and fix problems when they do occur.

(2) Become the market share leader of network file servers.

This goal follows from the first. Unless we lead the file server market, we obviously have not convinced the industry that filers in general, and NetApp filers in particular, are the best solution for storing network data. (Right now we've got 1 to 2% percent market share in NFS. The leader, Sun, has 35%. Moving into the NT space will grow our target market and decrease our market share!)

Note that the goal is not to be the market share leader in filers. We accomplished that goal back in 1993 when we shipped the first NFS filer to Patrick Mulrooney at Tandem!

(3) Become a Fortune 1000 Company.

Please understand: GROWTH FOR THE SAKE OF BEING LARGE IS NOT A WORTHY GOAL.

We must grow to meet the objectives above. We cannot change the industry and lead the market as a small company.

Even though growth can sometimes be confusing, stressful, and painful, it can also be exciting, especially when growth is what allows us to make a real difference in the industry.

It should be obvious that James and I have not achieved our goals. From where we started, we've come less than half way. The challenge ahead is at least as large as the challenge from founding to IPO.

What we have accomplished is to build a foundation for the real mission. The people here now will, over time, become the core that holds NetApp together as we grow.

Thank you all so much for joining us!

James and Dave

Just to put things in perspective, let me point out that we had about 100 employees at that time, and maybe $50m in revenue per year, so becoming a Fortune 1000 company was an aggressive goal.

NetApp Passed EMC to Become the Largest Networked Storage Vendor (by Terabytes Shipped)

Last quarter was a great milestone for NetApp. We just passed EMC in terabytes shipped, to become the largest networked storage vendor. (I define networked storage as SAN, NAS and iSCSI for everything except mainframes.) Here are IDC's numbers for the seven largest vendors:

68,898 NetApp
66,581 EMC
61,050 HP
31,648 IBM
20,835 Hitachi
19,651 Dell
18,475 Sun

In terabytes shipped, the "big three" are pretty close together, but there is a big gap between them and #4. IBM is a factor of two lower. There is another factor of two drop to the next vendor below Sun, so you can think of the storage market as the "big three", the "medium four", and the "little guys".

Our jump to #1 in terabytes shipped creates a very interesting situation: There are three "first place vendors" in networked storage. HP is largest by units shipped, NetApp is largest by terabytes shipped, and EMC is largest by dollars collected. Analysts usually focus on dollars, which makes sense for understanding the market today, but to understand the strategic potential going forward, you can make a strong case that units and terabyte are both more important. You could argue that units are most important because that measures how many different installations you are helping customers with. Having more small units also positions you for a "disruptive attack from below." You could argue that terabytes are most important, because that measures how much customer data you are holding, and it represents the opportunity that you have to sell additional functionality. But in the long run, I think it's hard to argue that charging more money for less stuff is the best position to be in strategically.

My last blog on market share observed that for 2005, NetApp had 8.0% share by dollars, but 14.9% share by terabytes. Last quarter, we had 9.5% by dollars and 21.3% by terabytes. I'd love to claim that the difference is entirely because NetApp offers more terabytes for the money, but the real answer is more complex. I've found at least three reasons, besides "good value", that NetApp looks cheaper:

1. IDC's report on storage software shows $150m in file system software revenue for NetApp in 2005, which is odd since other vendors like EMC and Hitachi show no such revenue at all. The revenue is protocol licenses for NFS, CIFS and SAN. Other vendors bundle these protocols with their storage systems, but since NetApp has "unified storage" that supports all protocols in one box, we sell the licenses separately. Adding this in would bump NetApp's revenue by about 15%.
   
   (Since I discovered this, I've talked with the folks at IDC about moving this revenue from storage software to storage systems, since the current approach makes apples-to-apples comparisons difficult, but I'm not sure yet how that will turn out.)
   
   
2. Different vendors bundle different levels of support with their storage systems. For instance, EMC bundles three years of "premium support" with their systems, while NetApp bundles three years of "standard support". For EMC, I think the extra revenue for premium goes into storage systems, since it's bundled, but for NetApp into storage services. On the other hand, I understand that EMC is about to change the support bundle for CLARiiON from "premium" to "standard". I haven't figured out anything clever to recommend to IDC here.
   
   
3. Although I have no hard data, I believe that NetApp sells a much higher percentage of ATA disks than other vendors. We started early in this area, and with RAID-DP to protect against double disk failures, we have allowed customers to use ATA drives safely in a much wider range of applications than other vendors. Using ATA drives allows us to sell more terabytes for fewer dollars.

Summary: NetApp sells much more storage than it would appear based on revenue numbers.

That's enough geeking out on numbers for the time being.

[Quarterly numbers are from IDC's "Worldwide Quarterly Disk Storage Systems Tracker, Q2 2006 Release", August 2006. Annual storage system numbers are from IDC's "Worldwide Disk Storage Systems 2006-2010 Forecast and Analysis", May 2006. Storage software numbers are from IDC's "Worldwide Storage Software 2006-2010 Forecast", May 2006.]