Linked by Eugenia Loli on Tue 19th Nov 2002 09:24 UTC
Oracle and SUN This past year has been a breakthrough for Sun for both their Linux and Solaris products. The most intriguing news of all is possibly the challenge Sun poses to Microsoft with their Desktop Initiative announced a couple of months ago. We spoke to Bill Moffitt, Product Line Manager of the Solaris Lifecycle, about Linux, the desktop and Solaris. Update: Bill Moffitt replies on our forums.
Permalink for comment
To read all comments associated with this story, please click here.
Last comment...
by Bill Moffitt on Thu 21st Nov 2002 01:23 UTC

Red Pill, your rants on this board clearly indicate your biases. You distort your view of the world by bending fact to meet your opinion. In the harsh light of day your arguments just don't hold up.

As you cite no source, I shall consider the points you state as "fact" to be opinions until substantiated. That said, I mostly agree with some of them, particularly:

- Linux shipments are growing at 35% a year
- Linux system revenues growing at 26% a year
- by 2Q03, all server segments will exceed 4GB RAM capacity
- in 2003, industry standard processors will capture MPU performance lead
- low-end 64 bit systems will be commoditized

I don't know if the numbers are right for Linux shipments and revenue, but they sound about right. Linux is growing fast because it's good, it's cheap, and Linux+x86 really works for desktops and low-end servers. Additionally, the current economic client strongly favors low-end machines, which favors Linux. I don't know if 2Q03 will be when all server segments will exceed 4 GB of RAM, but I know that whenever the software *you* are running on *your* server requires more than 2 GB you have to move to a 64-bit platform, which doesn't currently favor Linux. On the last two points, however, I quibble only with tense: industry standard processors (meaning x86, MIPS, Alpha, PowerPC, PA-RISC, and, of course, SPARC) captured the performance lead a long, long time ago, putting "special purpose" processors out of business. (maybe you had something else in mind when you said "industry standard processors," so you might state your assumptions). Also, low-end 64-bit systems were, to a great degree, commoditized a long time ago, as well. You can buy a brand-new, fully-functional 64-bit 1P1U system today for $995 (Source: SunStore, SunFire V100). Yep, that's about $200 more than a low-end 32-bit x86 system (Dell PowerEdge 350, on sale starting at $799 with a whopping 128 MB of RAM, according to their web site), but, fully-configured, the prices are so close there isn't really a "64-bit" or a "Sun" premium. I only know of one vendor selling 64-bit systems in that price range, though - Sun. Apples-to-apples, the argument that Sun's stuff is way more expensive just doesn't hold water.

Other unsubstantiated "facts" you state are:

- 95% of all server systems sell at below $25K
- x86 dominates low-end server revenues
- systems priced above $25K generate 65% of all server revenue

Since you don't state a source, I don't know whether these numbers are valid or not. In any event, they're OK with us, because we're the only folks out there selling a whole line of 64-bit servers for under $25k, as well as the LX50 x86 server that can run either Solaris or Linux, as well as a line of (yes, higher-margin) servers for over $25k. It's kinda hard to see how this data, even if it is fact, hurts Sun's viability.

Finally, however, you sneak in one "fact" that's just plain not true:

- SPARC unit share is rapidly declining

IDC and Gartner/Dataquest numbers, as recently as their 2QCY02 reports, indicate exactly the opposite: SPARC continues to gain market share, especially in the entry-level market. Please feel free to refute them and cite your sources.

Given the shaky nature of your assumptions, it's pretty easy to refute the conclusions. In order:

"1. There is no long term future for SPARC."

SPARC is today the volume-leading 64-bit chip architecture in the market and the low-price competitor in that market, and we're investing aggressively to ensure that remains the case. Alpha's dead, PA-RISC is dying, MIPS is an embedded processor. Power4 is currently the high-price, low-volume alternative. Itanium is a market failure (and there's no evidence Itanium2 will fare any better), and we have yet to see what Opteron will be. Today, however, the 64-bit market is SPARC's to lose.

"2. Linux is expanding far more rapidly than Solaris. In simple terms, Solaris on SPARC has no long term future except on Sun's high-end big-iron systems."

I'll concede the first point, but it doesn't naturally lead to the second. Once again, Linux and Solaris are very different critters, and Linux's gains in the 32-bit space are much more likely to hurt Microsoft than Sun. Solaris on SPARC has a brilliant future as the need for a 64-bit architecture (per your own observation about the need for larger memory space) moves down into the low end of the market.

"3. Intel/AMD are going to enable many companies to even more ship Sun-killer servers, most of them powered by Linux."

Well, since we (Sun) are building a line of x86-based servers running either Solaris or Linux, it's kind of hard to see them as "Sun-killers." Linux on x86 is a great solution (I'm writing this message on my Sony Vaio running Red Hat 8 - you don't have to sell me on Linux!), but it's not the extremely stable, highly scalable, 64-bit desktop-to-mainframe solution that Solaris on SPARC is. The "competition" between Linux and Solaris is FUD, planted by companies who don't have a market-leading 64-bit product line. Clear analysis of the real condition of the market and what people really buy servers for leads inexorably to this conclusion. Don't believe the analysts (you don't know who pays them), don't believe me (you know who pays me) - strip away the crap and look at what's real.

By the way, who pays you? I'd sure be interested in knowing.

"4. Sun has no sustainable competitive advantage in servers."

Both true and not true. True - we don't have a magic widget that allows our servers to run faster, quieter, cooler, or perform feats of prestidigitation. Our servers are well-engineered machines built on a set of open standards (like SPARC) with "proprietary" features built on top to add value (like Dynamic Reconfiguration). Solaris is a well-engineered OS built on public standards with extensions to take advantage of the hardware upon which it runs (again, like Dynamic Reconfiguration on Sun hardware or PXE on Intel hardware) and specific features to make it less expensive overall (like Solaris Containers). However, at the same time, our real sustainable competitive advantage is our ability to put those pieces together into high-quality products that work at least as well as others at a very competitive price.

Embracing Linux and x86 is just an extension of this - our box is no different from Dell's, and our Linux is no different from Red Hat's. We're integrating the whole Sun ONE stack on top of that to bring the J2EE architecture to the low end server space, and backing it up with Sun service. Pricing is competitive with everyone else out there, so you're right, it's not a high-margin business, but we're happy to make money on volume, on add-on products, and on service.