Phasor Burn

Warning: Do not look into phasor with remaining eye.

About

Yet another collection of random links and rantings of a greying unix geek with a photography bent. Pass the Guinness and Grecian Formula.

Sun tries to give the appearance of being both a hardware company and a software company. That is a lie, however, since Sun is giving a lot of it’s software away for free (as in beer) now. That’s ok I suppose, if it’s used to drive hardware sales.

In other words, they are primarily a hardware company. Similar to Apple in that respect. Except Apple is rising in (desktop) market share, and Sun is declining (server market share).

Despite many years of trying to make inroads on the corporate desktop with thin clients and the misleadingly named “java desktop”, Sun really is all about servers. Servers mainly in heavy lifting roles (database) or internet facing (web server).

After the dotcom bubble 1.0 burst, there was such a glut of their hardware available on Ebay, as investors tried to get some pennies-on-the-dollar out of all those failed online petfood and similar ventures. Couple this with the rise of Linux plus the heavy backing of Linux by IBM, and Sun was hurting for hardware sales for many years. I don’t think they ever have recovered to anywhere near the pre-bubble sales even.


Fast forward to late 2005.

Sun now has some x86 based products, which can run Windows, Solaris, or Linux as supported configurations. Sun seems to be flailing around looking for something that will ‘catch’ that they can then use start leveraging themselves out of their money-pit. Yes, they have thrown themselves in with Microsoft of all things. Anyways, these are really first generation servers, and my experience with them thus far is that they are a bit half-baked and not as solid as some of the sparc gear can be.

In particular the goofball service processor on the V20z and V40z can’t hold a candle to the basic ALOM available on the lowest of low end sparc systems. I was unable to successfully install a base os on those two models without using a physical KVM attached to a real keyboard and monitor. The serial bios interface wasn’t quite all there and capable of performing remote os installs,even with the media in the local dvd drive. Performance is ok, they did seem to get some reasonably solid engineering done on memory interconnects and integrated raid controllers etc.

Fast forward to 2006

Many if not all datacentres are changing their pricing structure away from bandwidth and space to “how much power does your server use”. Afterall, the cost of oil for power generation has gone up as well as all the cooling that has to happen to get rid of the heat these systems make. You can only go so far with faster, hotter, louder, 1U, 2U, 5U server behemoths.

Performance per watt of power is fast becoming the new measuring stick.

Apple is doing it with their move from PowerPC G5 to Intel Core Duo. Sun is doing it with their new T1000 and T2000 “Niagra” product line that run 8 cores per cpu package, 4 parallel threads per core = 32 threads per cpu package. This has been a few years in the making and is now just becoming commerically available.

Seriously ‘cool’ stuff. Pun intended.

As part of a promotional effort, Sun is attempting to loan T2000’s at no cost for 60 days to prospective customers. They even have a program where if you benchmark the unit against other systems and write up a good report (positive or negative) in your blog, that you may get to keep the loaner T2000 indefinitely. Example blog benchmark posting.

This sounds almost desperate enough to work.

However in the comments section of Schwartz’s blog entry about this loaner / benchmark / blog it / keep it program there are many people saying how hard it is to get approved, to even get the time of day from Sun’s sales people.

Now, I R not a business person. I’m just a systems administrator who’s seen a solid hardware company fall after flying too close to the, er, sun. I sure would like them to turn it around and become successful again, if only to have something other than Dell and HP as options, and as the last “real unix hardware” widely available. (SGI doesn’t count, IBM’s really not pushing their AIX gear very much).

But i really don’t think Sun as a company is going to survive, no matter how ‘cool’ their new hardware is. Not with the problem in their sales department. If it doesn’t get addressed, they’re dead but just don’t know it yet.

This comment on Jonathan Schwartz’s blog in particular hits the nail right on the head.

You are a global company, selling globally. Until anyone in the world can go to your web site and order a box with a credit card, and have it delivered 2-5 business days later, Dell and others are going to keep trouncing you. You will notice that there are no salespeople involved in that transaction.You will also notice that there are no sales territories, no quotas, no “you must go through your rep” stories in my scenario.

Your sales practices are actively hindering adoption of your products.

It’s just too sad, really. Worse, even, than when the IT department’s rules and processes and what not actively hinder a company from doing business. (Another common thing with large technology companies, but that is another ramble/rant).

For independent confirmation of how hard it is to buy servers from Sun, check out The Sun doesn’t shine on me.
Sigh.

Closely related to sale, but distinctly different, is marketing. Zoltar knows, Sun has given us confused marketing messages over the years. x86 bad, x86 good. Linux bad, Linux good, Linux bad. Solaris x86 yes, Solaris x86 no, Solaris x86 yes.

Dilbert - vp marketing

All good stereo types come from some part of reality. Maybe this explains it all.

Leave a Reply