First, EMC and NetApp are CURRENTLY direct competitors in many situations. This is not an eventual, this is a now. Granted, NetApps only has one card to play (low-end NAS) so it will not come up against EMC in every sales situation. However, in the growing NAS market, EMC and NetApps are seeing more and more of each other.
NTAP has several cards to play, including its ability to build a end-to-end storage and data replication/distribution solution using filers and multi-media caching appliances built on the same hardware/OS architecture. As for the "low-end NAS" remark--NTAP's capacity ranges from less than 30GB for single server configs to 6TB for dual clustered configs. As we have seen with recent wins and heard from DW's comments on the most recent cc, gauging NTAP's scalability by the maximum disk capacity of a single filer is not really valid. An enterprise can utilize dozens of filers quite easily giving 100's of TB and administer them in a centralized or a distributed fashion. (Deutsche Telecom, for example:)
netapp.com
Regarding the statement that, as LAN speeds grow, NetApps is positioned more and more strongly: LAN and WAN speeds are without doubt growing exponentially. With technologies like DWDM (Dense Wave Division Multiplexing), we are beginning to realize that we won't soon hit a limit on the capabilities of fiber network bandwidth. Further, Ethernet networks are BEGINNING to move to a point where, from a pure bandwidth perspective, they approach the throughput of direct channel attached systems.
That said, it's not much of a logical leap to figure that part of the 2 billion EMC is spending in R&D over the next two years is being spent in this arena (note strong engineering partnerships with Nortel, Lucent, and Cisco for proof of this). There are definite hurdles to overcome before Ethernet (NAS) can challenge channel (SAN) architectures as the de facto storage solution. It may never happen. That is yet to be decided. It is definitely not a "this year" kind of thing. It is a bit out there before it has a real bearing on product offerings and market standing, if ever.
Who says that SAN is "the de facto storage solution"? The fact is, the de facto storage solution is still server attached disk arrays from the computer manufacturer (SUNW, HWP, IBM, etc.) They still have over 75% of the storage market!
The author is concentrating on the ethernet vs SAN issues. The storage network business is exploding with new products coming to market including GbE and InfiniBand. So much of the discussion about "NAS vs SAN" revolves around this ethernet vs fibre channel (FC) debate. The fact is NTAP accesses its disk arrays via (FC). Ethernet (or FDDI, or ATM) is the media attaching the application host computers to the filer. (Application hosts include desktop PCs, Unix servers, NT servers, etc.) The cable segments attaching these hosts to the filer can be configured to serve the data loads appropriately. For example, if the objective is to consolidate data from hundreds or thousands of desktops, then the filers reside on the corporate LAN. If the objective is to serve a high volume RDBMS, then the filer(s) are connected to the RDBMS app server (the Oracle host) completely isolated from the client PCs (corporate LAN).
Future technologies, such as DFAS/VI will take this argument away and provide NTAP with data transfer throughput equal to EMC's channel attached architecture with more flexibility and lower cost. VI will allow filers to communicate with hosts and with one another without the burden of the TCP/IP stack and packetizing. VI will be a leap ahead of direct channel connect at a much smaller price.
At any rate, when speaking of future technologies, which is what the article is doing here -- the real question is: who will have a solution that best leverages the emerging technology? My guess is it is the company who has partnered with the best, and put the most resources forth. Because of the financial capital that EMC can bring to bear, and the engineering partnerships it forms, EMC has consistently been the first to market with items of this nature in the storage arena. I don't see this changing. If the market goes to Ethernet based storage, EMC will not have it's pants down, wondering what happened.
Don Valentine (ex Chairman of CSCO, current board member of CSCO) and the CFO of CSCO are both members of NTAP's board. Do you think NTAP has a strong alliance with CSCO?
NTAP has founded and participates in every major storage-related alliance promoting open standards. I don't think EMC has any advantage here.
In fact, EMC must spend large parts of its resources supporting its installed base and providing new upgrade products for that base. What money they are spending on NAS is not known by the public.
It is also interesting that folks speak of EMC as a late comer to the NAS market. EMC has had a NAS solution for years -- quite frankly they were slower to notice the market begin to accelerate. NetApps definitely got in under EMC's radar and grabbed a good amount of market and mind share. Without a doubt NetApps, as a company, is both very fast and very determined.
A page out of Christensen's book, The Innovators' Dilemma. Heretofore it has not been in EMC's interest to have a NAS product. Even now, their NAS product is just a network front-end to the EMC Symmetrix SAN product. It remains to be seen if EMC is the exception to the usual response to disruptive innovation. (Read the book.)
However, I don't think anyone would argue that the 800 lb. gorilla has now turned his head, and put a strong focus on marketing a mature NAS solution. See the corporate web site (SAN vs. NAS debate)today if you have any doubt.
EMC is not a Gorilla. It is big, but it does not command the storage market. Also, will EMC strive to succeed in the NAS market or strive to kill that market. And most importantly, NTAP's disruptive innovation is NOT based on NAS. It is based on its proprietary file system (WAFL) which dis-integrates the UNIX and Windows file system, reducing the real bottleneck to file system performance and allows them to continue the evolution of that innovation with such things as SNAPSHOT, DAFS/VI, and soon-to-come new clustered configurations. WAFL also reduces total cost of ownership by simplifying the sys admin of an enterprise's storage assets.
To be clear, EMC has in no way dis-integrated the native file systems (UNIX/Windows) from the storage system. They have merely replicated it inside the SAN. NTAP has removed it from the storage subsystem.
In turn, EMC's market share is growning faster in the NAS arena than NetApps is (i.e. if the growth rates were to continue, EMC would surpass NetApps in the near future). Lest someone else point it out, it is obviously easier to grow market share from 10% to 20% than it is from 40% to 80%. so obviously EMC is growing faster percentage-wise now.
But the point here is that EMC is indeed focusing on this area very strongly. And contrary to what I've seen in other posts, for EMC to focus on NAS does not cannibalize it's margins elsewhere. I think what many people are still tending to miss on this topic is this very key point: NAS is COMPLEMENTARY to SAN, not a REPLACEMENT for it.
Should I even bother to address the "fastest growing NAS market share" claim? Let's not forget that what EMC calls a NAS product is the entire configuration of its NAS front end with a Symmetrix SAN.
EMC is not cannibilizing its margins because its approach is not to offer a more cost effective NAS solution, but to front end its high margin Symmetrix with a NAS front end.
The distinction between NAS is SAN is already so foggy that it is meaningless. What customers will be looking at for storage solutions will be flexibility in implementation strategies (distributed and centralized), scalability, reliability, security, and total cost of ownership. NTAP will perform very well against EMC based on these criteria.
For instance database technologies (Oracle, SQL, etc.) require channel speeds and consistency (SAN).
What database technologies require depends on the customer's implementation of that infrastructure. Substitute the word "throughput" for channel speed. Channel speed is EMC's approach to throughput. NTAP's approach is based on the efficiencies of WAFL and its unique ability to dis-integrate the RDBMS server from the RDBMS data server. NTAP does very well in this market and is the ONLY platform supported by ORCL, Informix, and Sybase for NFS attached storage of RDBMS data. This gives incredible options to the customers that no one else can offer.
Other things like web-content delivery and replication fit better in the NAS space. What is often needed is a solution that can accomodate both needs.
Agreed.
At the end of the day though, the age old questions on the Fool boards have been and I think will continue to be:
1) Who is going to hurt who the most here?
2) As these two companies take aim at each other, is it safe to be long on both stocks?
The facts bear out that NetApps jumped on this market like a pit bull and has no intention of letting go. The facts also bear out that EMC is currently putting a very concerted effort to, not only grab share in this market, but to depose NetApps and become the market leader in it.
So, if you think EMC will win, IMHO you can hold on to NetApps for about 3 quarters more, maybe a year, before it becomes a dangerous holding. If EMC's efforts in this arena are successful, it won't take much longer than that for the market to see the trend, and you need only look to Auspex to see what happens when a company with a singular focus in NAS loses it's market share.
If you think EMC will win, sell your NTAP right now! Don't wait!
Auspex has a closed, proprietary architecture based on custom manufactured communications interfaces and a modified version of UNIX for the OS. Their failure was not because they had a "singular focus on NAS". Their falure was because their total cost of ownership was eclipsed by NTAP's off-the-shelf hardware component architecture, simplicity of operation with the ONTAP OS, and performance/reliability advantage of WAFL. They also failed because NTAP recognized that the best way to grow a customer base was to go after AUSPEX customers who were facing expensive upgrades but understood what NAS was already. (It was actually less expensive to buy a new NTAP filer than it was to upgrade the AUSPEX system to support the latest ethernet speeds and add storage capacity.)
Guess who else has a closed proprietary architecture.
NTAP will not, however, focus on EMC's customer base. They will focus on beating EMC for new business, however.
On the other hand, if you think NetApps will win, I think you can continue to hold both NetApps and EMC long. Even if EMC loses this battle on NAS (I don't see this happening by the way), it remains more than viable on too many other fronts.
The more like scenario and I agree completely.
Disclosure: Long NTAP. Was long EMC for 2 years, but sold when they bought DGN and hired Tucci for President. Will be holding NTAP for several more years and consider it one of my safest investments for the long term (3+ more years). |
