NetApp - Dave's Blog

Data Retention Policy: US Lawyers versus UK Lawyers

Many companies are struggling to figure out whether it's best to delete e-mail, and other potentially incriminating data as soon as possible, or whether it's better to keep it forever.

The "as soon as possible" camp argues that saved data is simply bad news. The more data you keep, the more likely that something incriminating -- or at least troublesome -- will be found during discovery for a legal case, and then you will wish that you had deleted it. If you are legally required to keep data for three years, then delete it after three years and one day.

The "keep it forever" camp argues that there are so many copies of data that it's almost impossible to delete them all. With e-mail, for instance, both the sender and receiver have a copy, maybe many receivers, and each of them may have saved it in lots of different folders, or even sent it on to additional people. And don't forget backups. The "keep it forever" people argue that any incriminating message is likely to be found eventually, and you are better off finding it yourself as soon as possible, so your opponent doesn't spring it on you at the end of the trial, where it looks like a smoking gun you were trying to hide.

We recently had a conference in Europe several of our top European customers, one of whom ran IT for a large law firm in the UK. He had an interesting perspective. In the US, he said, lawyers all want to delete their data as quickly as possible. They seem to assume that any data they keep is likely to incriminate them. In the UK the lawyers are the opposite. They all want to keep their data as long as possible. They seem to assume that any data they keep is likely to prove their innocence.

Never mind the motivation, my sense from talking with a variety of customers is that, at least for unstructured data like e-mail, people are increasingly leaning towards the "keep it forever" camp.

Hard Problems, Gross Margins and Smart People

NetApp consistently has gross margins of 60-62%. That's been true for about ten years now, ever since we went public. Our financial analysts have always questioned whether we'll be able to keep gross margins high.

For people without a finance background, gross margins are the "mark-up" that a company adds to the cost of manufacturing its products before passing them through to the customer. A 60% GM means that for every $100 NetApp charges a customer, we spend $40 manufacturing our product, and $60 goes to other things, like salaries, equipment, rent, taxes and profit. Dell has 20% GM, so for every $100 from the customer, $80 goes into manufacturing the product, and only $20 goes to other things.

People ask lots of questions about high gross margin companies. Are you ripping off your customers? Aren't you afraid that your business will commoditize, and your gross margins will fall? Won't Dell win in the end, because they've figured out how to be profitable at 20% GM - that is they've figured out how to have a much lower mark-up than you?

The answer to all of these questions is "No." To understand why, you have to look at how we spend that $60. It's not like it all goes straight to profit. We spend it largely on hiring smart people and building the environment for them to do their job.

In essence, being a high GM company means that the customers have really hard problems that they are willing to pay someone to solve.

When you ask whether the gross margins are going to fall (will the business commoditize), the real question you are asking is whether the customer is either going to stop having really hard problems or stop being willing to pay money to have them solved.

In storage, I see no end of hard problems. I see customers' data growing, and I see customers worried about consolidating, operating and protecting increasing amounts of data. I see the burden of ownership for data getting worse as businesses rely more and more on the data they own, and as the legal requirements increase to keep data accessible yet private. No end of problems in sight.

Dell is an interesting case. In some businesses, a low GM means that there are no hard problems to solve. A grocery store has a low GM, and that just means that meat and vegetables are "well understood" - there's little value anyone can add. In Dell's case, there are hard problems involved in making PCs faster and better every year; it's just that Dell outsources the solving of them. Much of Dell's manufacturing costs go to companies like Microsoft, Intel, or - for storage - EMC, all of which have much higher GMs than Dell. The difference between Dell and a grocery store is that for a grocery store it is low margins all the way down, but for Dell, the high margins and smart problem solvers are one layer back. (I should say, there are plenty of smart people at Dell. It's just that they focus on improving the manufacturing process itself, lowering costs there, and not on improving the products or solving customer problems.)

When Dan Warmenhoven, CEO of NetApp, says that he doesn't worry about commoditization of our business, he doesn't mean that there will be no commoditization at all. There are certainly parts of storage, like basic RAID shelves, that are commoditizing, and I'm sure there will be more. What he means is that customers still have plenty of hard problems they want solved. Good news for gross margins.

EMC's Quantum Leap—Expanding Beyond Storage

You can learn a lot about a company's strategy by looking at the markets it is expanding into. In my last posting, I looked at NetApp's current "quantum leap" from storage systems (SAN, iSCSI, NAS) into capabilities like encryption that add value to storage from other vendors.

EMC is also taking a quantum leap, exemplified by the acquisition of VMware and SMARTS. This expansion goes completely beyond the storage market, which will have very interesting strategic implications, both for EMC and for others in the IT business. As in my NetApp analysis, I'll look back at EMC's earlier quantum leaps, and then return to the present.

EMC started in mainframe storage where they beat IBM DASD by building a very cost effective solution based on commodity SCSI disks. (A classic disruptive strategy—SCSI started as a low-end disk for workstations, and EMC took it into the enterprise.)

EMC's first quantum leap was to expand into open systems storage. Not only did they support UNIX and Windows with Symmetrix, but they also acquired CLARiiON to move down market and acquired CrosStor to enter the NAS market. (Before this expansion, EMC and NetApp competed little, but EMC's expansion combined with NetApp's entry into the SAN market put the two into head-to-head competition.)

EMC's next quantum leap was into storage software and ILM (Information Lifecycle Management). The acquisition of Legato exemplifies this expansion, and also BMC's storage management software.

Today EMC is expanding completely beyond storage, although they don't yet position themselves that way.

The first step was Documentum. EMC positions Documentum as part of ILM, but I don't buy it. EMC could equally well have bought Sybase and positioned it as ILM because a database creates and manages records. To me, Documentum is a step outside of ILM and storage—or at least a half-step.

EMC's more recent acquisitions of SMARTS and VMware make this trend even clearer. SMARTS takes EMC into data center management beyond storage, but VMware is where the storage link breaks down completely. VMware is not about storage, it is about owning the application environment.

I have a hunch that this explains IBM's increased aggression towards EMC. EMC has competed against IBM over storage for years, but now EMC is starting to compete for control of the application and server environment. For a computer company like IBM, that is much closer to home.

Microsoft also wants to own the application and server environment, and MS Virtual Server for Windows competes directly against VMware, so EMC's move could create tension here as well.

To be clear, I think that EMC's strategy is challenging, but I also think it may be the best strategy available to them. Because of their maturity in enterprise storage and data management, expanding beyond these areas may be their only hope of maintaining the growth rate that Wall Street demands.

In closing, let me take the opportunity to clarify NetApp's strategy by contrasting it with EMC's. NetApp is still small enough to double entirely within data management and storage, to double again, and still have room to grow more. Someday I would love to have EMC's problem, and be forced to expand beyond the data and storage. And I certainly admit that scale and breadth can bring some advantages. On the other hand, NetApp gets to stay focused on simplifying data management, and focus creates a variety of advantages for a company, for its partners, and for its customers.

In strategy, you always have to play the cards in your hand. Today, EMC and NetApp have very different hands.

Network Appliance's Quantum Leaps

Any healthy, growing company broadens its product line and its customer base over time. It moves up market or down market or adds features to better address new verticals. NetApp is always doing that kind of incremental expansion. But three times now - in our history - we have made quantum leaps to our product line that dramatically broadened the market we serve.

Our acquisitions of Decru (data security through encryption) and Alacritus (disk-based data protection through virtual tape libraries) are part of the third quantum leap. To clarify our strategy, let me first describe the first two quantum leaps, then I'll come back to the third.

NetApp started as a UNIX NAS company. Our first quantum leap was when we expanded ONTAP to support both UNIX and Windows. Today, this seems like a no-brainer, and people expect every NAS product to serve both, but at the time there was a deep gulf between UNIX and Windows. It was a big leap for customers to accept that the same product could be the best Windows NAS as well as the best UNIX NAS. It was a big cultural shift for NetApp employees, as well, since the early ones tended to be UNIX bigots who hated Windows.

The second quantum leap was our expansion into block-based storage - both Fibre Channel and iSCSI. Again it was a big cultural shift, because many of our employees at the time were NAS bigots who argued "NAS is good and SAN is bad." I was the ring-leader, and it took James over a year to help me see the light. Then I spent years converting others. That was almost five years ago, and since then we've moved to the upper right of Gartner's magic quadrant for SAN, farther to the up and right than everyone else except EMC.

The third and current quantum leap is to broaden into heterogeneous data management - to expand beyond features that operate only on NetApp storage and offer products that support non-NetApp storage as well. The Decru Datafort is a perfect example. Sure it'll encrypt NetApp storage, but it'll encrypt EMC, HP, Sun or IBM storage just as well - even StorageTek tapes. Likewise, the VTL (virtual tape library) from Alacritus will support many vendors' storage.

This leap again requires a cultural shift. To be successful, NetApp must partner with storage vendors that we've traditionally competed against. Being buddies with EMC is a mind-shift for NetApp employees on par with embracing Windows back in 1995 or embracing SAN back in 2000. Some have asked, "Why would we give EMC access to Decru's cool technology? Why not lock them out?" The market for Decru is much larger if we can support all storage vendors - that's why. NetApp has about 10% market share in open-system networked storage, so that's another 90% of the market we can go after if we keep Decru heterogeneous. Today, Decru is both an EMC Select Partner and an EMC NAS Partner.

The recent restructuring of our product operations under Tom Georgens and Jay Kidd reflects and supports this new strategy.

Tom G. runs Enterprise Storage Systems, which is our ONTAP-based SAN, iSCSI and NAS storage systems along with all of the homogeneous data management services that support it, like SnapMirror, SnapVault and Data Fabric Manager.

Jay owns all of the heterogeneous data management product lines that support non-NetApp storage as well as NetApp storage. In the same way that EMC structured itself to "wall off" VMware, to allow it to cooperate with EMC's traditional competitors, we are structuring to wall off Decru and Alacritus.

In some ways this is our biggest quantum leap ever. In the past, we always added new capabilities to ONTAP. Now we are broadening our product line beyond ONTAP. In the past we expanded to add new competitors, while keeping existing ones, but now we are converting competitors to partners. (Decru is a reseller partner with EMC.)

I think it's appropriate for a bigger, older NetApp to make bigger leaps.