Best of breed hardly cuts it for a company that says innovation is in its DNA. Across the technology board Sun builds better mousetraps: faster processors; containerised data centres; an extended zettabyte file system (ZFS); and more. Yet with its hard drive-based storage it doesn't seem to be doing this.
Looking across Sun's product lines we see enormous innovation: CPU chips with Niagara 2; software with xVM virtualisation and ZFS; and tape automation with the coming 1TB+ T10K drives. But then we look at hard drive-based storage and the theme is best of breed; innovation being restricted to apparent niche oddities, like Thumper.
With hard drive-based arrays Sun is seemingly content to sell its line of NAS products and SAN storage arrays, the Engenio-based modular mid-range products and the high-end Hitachi-supplied products, and keep pace with instead of outmatching its competition
Disk storage sales are ho, hum and declining
An IDC report on external disk product sales showed Sun and NetApp sharing the number six spot. Gartner has recorded a decline in Sun's disk array sales. Its sales in the external controller-based disk storage market, from Q1 06 to Q1 07, showed a 15 percent decline according to Gartner.
This is not a great performance.
Put crudely, Sun is selling NAS and SAN array products which match or exceed competing products technically in several ways, but which are perceived by markets and analysts as not doing well enough because of the sales numbers vis a vis the competition.
Yet Sun is not developing better SAN arrays or NAS filers in the existing mode at all. Where are the Sun-develped products to take on Isilon (clustered NAS), EqualLogic and Left Hand Networks (iSCSI SANs), NetApp and EMC (filers and SAN arrays)?
They don't exist and Sun appears to be content with its Engenio and HDS-supplied 'best of breed' technology for these markets. Isn't that odd? In fact this is a holding strategy. Sun is not about trying to outsell the competitors above in these traditional storage markets. It aims to both bypass and disrupt them and their business models.
How Sun is going about this with its disk-based storage products is utterly dependent on Solaris and what has happened to it.
The Solaris model
A few years ago Solaris was, roughly speaking, just another proprietary Unix facing prolonged decline under the twin assaults of Linux and Windows, plus sideways attacks from HP-UX and AIX. Today it is one of the most widely-distributed and used operating systems on the planet with razor-sharp, leading edge technology such as ZFS.
What happened? How did this come about? Sun saw that people, businesses, were not going to buy Solaris and Solaris-based systems in the numbers needed to make Sun's revenue goals. Solaris was, technically speaking, a very good O/S but, in a market sense it was a weak product, having no good answer to the open source Linux surge and not distinguished enough from HP-UX and AIX.
Sun turned this weakness into a strength by making Solaris free and selling support contracts for it. Solaris is turning into a distribution, like Linux. Developers anywhere and everywhere downloaded millions of free copies. The calculation is that some of these will be employed by business and some these businesses will buy Solaris-based kit from Sun. And so they are, slowly but surely.
Solaris is not a profit centre in itself. It is an enabler of support revenues and hardware sales. The (Solaris) volume drives value (through support and hardware sales) elsewhere in Sun.
Now Sun is building a better O/S mousetrap but giving it away in exchange for Solaris users' influence on business purchases of Sun products and support down the line. It is indirect and it is disruptive of the traditional UNIX O/S business model as employed by HP (HP-UX) and IBM (AIX).
What is the disk storage strategy?
Sun believes, passionately believes, that general-purpose commodity hardware and system software is more than powerful enough to out-perform specialised and proprietary hardware and software. (It counts SPARC as general-purpose and commodity hardware.)
Instead of using proprietary server/controller hardware, RAID technology and specialised, embedded storage system software stacks driving high-performance drives to produce iSCSI SANs, Fibre Channel SAN arrays or NAS boxes, Sun thinks that general purpose commodity hardware and software should be employed instead, to build products for high-growth storage markets. (See this blog note.)
Such software involves a 'thick' Solaris software stack including things like the ZFS file system, NFS, CIFS - for which Sun is building an open source version, and so on; system storage applications that are generally only available as proprietary software not unbundled from their containing O/S.
Sun says this thick software stack can run on SPARC, AMD and Intel-based hardware from multiple vendors, meaning no lock-in and lower prices.
Which storage markets should Sun address?
The SAN array market is mature and not growing fast; two percent a year perhaps, according to Jon Benson, Sun's SVP for storage. The NAS market is mature also. There are new and fast-growing storage application markets and these are characterised by vast, humungous amounts of relatively unstructured data which needs a lot of meta data processing to enable search, analysis and presentation in various formats of the data. Traditional SAN and NAS products are not good at this. Existing document management and business intelligence systems are either too slow or too high-end, specialised and expensive.
Here's the Sun thing; by building storage platforms aimed at sections of this new storage application market out of commodity processors, the thick Solaris-based system software stack and commodity drives, Sun can provide very powerful, capacious and cost-effective systems that can be customised by developers for their customers. These systems scale linearly and have balanced processing, I/O and storage capacities according to Sun.
Where does storage start and stop?
Such products defy traditional product classifications. Take the X4500 (Thumper). It is a server because it has AMD Opterons and RAM. It is software because it has the thick Solaris stack, and it is storage because it has lots of disk capacity and presents itself as a kind of super-NAS box. These are hybrid storage/servers. (It underpins why Sun recently merged its systems and storage engineering organisations.)
This provides problems for IDC and others because Sun's storage sales are much harder to tease out. Sun's storage strategy is, to this extent, necessarily opaque and, in the existing SAN-attach and NAS markets its showing is not likely to substantially alter in the near term. To that extent its storage strategy may appear weak.
But Sun takes great heart from Thumper's successful introduction, growing from nothing to a hundred million dollar run rate in just two quarters.
Thumper now has a NAS 'personality' and Sun can go into the general NAS market with a better NAS mousetrap, one with lower cost and higher performance than competing products. It has also added a virtual tape personality and turned Thumper into the VTL-V using FalconStor virtual tape code. A VTL-Ve, for extended, will come soon and that will include Falconstor's SIR de-duplication, making it even more attractive to buyers.
One piece of hardware and one set of software is being used in several ways, which lowers Sun's costs.
If this Thumper sales momentum can be maintained and if newer products like Honeycomb, aka the 5800, can replicate it and if - several biggish 'ifs' here - the tape product refreshers - Crimson (SL5800 downsized to mid-range), T10000B and 9840D - ramp up sales as they should - then Sun's storage sales will rise. Thumper and Honeycomb could collectively represent half a billion a year in sales by the end of 2008, possibly going past a billion a year or so later.
In theory now, any storage product category with specialised and proprietary and expensive controller hardware and software can be attacked by Sun using the commodity server hardware and thick Solaris stack approach.
Take EMC's Centera, which Sun people like Gail Truman, a senior product line manager at Sun, say is a first generation product, can be undercut in price, out performed by commodity hardware and system software by Honeycomb, a third generation product with intelligence incorporated directly with the storage.
Take NetApp NAS and FC/iSCSI SAN products; these can be out-performed at a much lower cost by X4500-based products. ZFS provides a global name space and obviates the need for any RAID controller kit. The Lustre acquisition means clustered X4500s will be able to provide Isilon and NetApp ONTAP GX-class clustering.
That lawsuit; put up or shut up
Sun CEO Jonathan Schwartz characterised NetApp's business model as proprietary software running on white boxes (arrays and server/controllers) with 70 percent gross margin. What's implicit in the Sun storage strategy is that NetApp will face tremendous disruption to its product strategy if Sun storage sales into the NetApp heartland ramp up.
Schwartz says the Thumper NAS is a third of the price of the near-equivalent NetApp product.
Sun people seem quietly convinced that this is what lies behind the NetApp IP infringement lawsuit against Sun. Sun was burned a few years ago and paid $100 million to Kodak in an IP infringement case. It is determined not to do this again, but, because ZFS is so central to its strategy, it is going to fight NetApp's case. Schartz said Sun had asked NetApp to reveal the infringed code but: "it hasn't done that." Then, in a quiet but quite distinct threat, he said: "We have 15,000 patents dealing with the Internet. That might be a risk to NetApp." In other words, put up or shut up and back off.
Bypass and then disrupt
What NetApp, EMC and other Sun storage competitors are seeing is Sun stepping back from focusing on initial product-for-product competition. It has by-passed this in the existing SAN and NAS markets and is maintaining a presence, a good presence, with products OEM'd from Engenio and HDS. But simultaneously Sun is preparing its own products, disruptive ones with lower costs and higher performance, built from commodity hardware and open source Solaris-based software stacks, that intentionally threaten to profoundly disrupt these competitors' business in the new and vast emerging market for storing relatively unstructured data.
Then, using these new products as a base on which to layer additional personalities, it is returning to the existing storage markets, NAS and VTL at first, and aiming to undercut and out-perform its competitors with a vast army of developers building products customers really want from these new integrated storage platforms.
That's the theory; bypass and then disrupt. Will it work?
Straws in the wind: IBM is to support Solaris; Microsoft and IBM are cross-supporting each other's products in the server virtualisation space; NetApp is suing Sun. Three very very large IT vendors reacting to Sun in their different ways and seeing impressive potential there; potential for good and potential for disruption. Sun is now full of promise for some and threat for others. Some it will thump, they fear, and others will earn golden honey from its hive of innovation. That's what Sun hopes, that innovation will feed its bottom line. We'll see.