Explorations

Is Sun Back?

Sun is often touted as a zombie company.  After all, they generally failed to monetize Java and their strategy depends on selling super-expensive closed systems, right?  That was then.  Sun has been plugging away for the last few years, making tweaks that finally seem to be paying off.  The major components of their strategy seem to be:

• Open-Sourcing Solaris and making it competitive to Linux with ZFS, Zones, etc.
• Niagra's way-parallel CPUs, with up to 64 thread units on a chip
• X4500's 24TB-in-4U density
• Java

It looks like Sun has a two-tiered strategy, where Java drives mindshare to Sun and hardware drives revenue.  But Java is an aging platform that is well into its maturity phase.  Java 1.5 didn't create half the buzz of (say) Rails 1.0.  So to create excitement, Sun has instead relied on its core asset: hardcore systems engineering.  Giving Solaris tangible advantages over its main competitor in the server market (Linux) was a smart tactical move.  But what I think will ultimately prove to be more valuable is the sheer value of these machines.  For instance, a 6-core T1000 starts under $4k and only takes 1U in a data center.  As a developer running this hardware, I'm already ahead even before any potential gains in power efficiency.  So far, it looks like the new product line is bearing fruit.

How did Sun accomplish what looks like the beginnings of a rise from the ashes of the dotcom collapse?  It may sound trite, but from the outside it looks like they just listened to their customers.  What did their customers want to do with their servers?  It turns out, not a whole lot of computation.  They want to serve lots of HTTP requests in parallel.  They want to push lots of I/O.  They want to store tons of stuff.  As Tim Bray says, Sun likes "to sell computers to run software on", a focus that should lead Sun to make Solaris the #1 platform for PHP, Rails, Python, J2EE, whatever. 

So Sun has damn sexy hardware and a fancy OS in Solaris.  So why aren't more startups running Sun?  I think the answer comes down to a hole in the Sun channel that is exacerbated by the new startup funding model.  In a nutshell, new Web companies don't usually have tons of money.  Many of the prominent firms were bootstrapped or ran on very small investments for much of their lives: del.icio.us, flickr, 37 Signals,Odeo, Blogger, Writely.  And that's not even considering the Y Combinator startups.  Even the firms that do raise capital (i.e. YouTube) are much more likely to have started in a kitchen than in a VC boardroom.  And for the purposes of launching their sites (there are no more 2-year site build-outs anymore), they are all essentially running on personal savings.  So they will all go through roughly the following process to get their first servers:

1. Google for "dedicated hosting".
2. Click a link.
3. Sign up the same day.

They will generally not colo because they either have other jobs, or don't want to pony up the money to buy redundant hardware, or would rather build the site than configure RAID all night. 

The problem for Sun is that following these 3 steps is almost guaranteed to NOT get you to a Solaris machine.  Googling for "dedicated hosting" gives ~92m hits, "solaris hosting" only ~7m hits.  And once you've started in your kitchen on Linux, it's going to take quite a bit of convincing to get you to port your infrastructure to Solaris.  Even a productions site running something like Rails+MySQL, will have tons of scripts that Just Work on RHEL or whatever Linux you chose.   It doesn't take a genius to realize that moving to Solaris is not going to be a zero-effort migration.  Plus, once the site is launched you have bigger things to worry about, like finding a business model. 

So Sun needs to catch startups at the time of their creation.  One way they could do this is by courting the big dedicated hosting companies (i.e. Rackspace, etc.) and getting them to offer Solaris.  Many companies would pay a small premium to get the Niagara machines, and we would kill for something like the X4500 over the crappy storage options Rackspace has.  If Sun was in the managed hosting centers, they would capture today's startups like they owned the Web 1.0 startups.  Hopefully, their evangelism will move down to the managed hosting companies so that we can get some X4500's.

So I generally like Sun's trend but I'm not sure why they are missing this obvious gestator of startups.

Dell Drops MP3 Players

According to the WSJ, Dell is exiting the MP3 player market.  This is a move that absolutely makes sense from a business perspective.  Dell's strengths lie primarily in delivering commodity products built around a standard platform cheaply by exploiting their excellence in supply chain management and just-in-time manufacturing.  But the portable media player is a poor overlap for these strengths from the beginning because it is a market that is:

• Not commoditized.  Consumers perceive a huge difference between the iPod and the iriver.  The winning featureset is still being sorted out by the marketplace.
• Affected by fashion.  Dell is not a "cool" brand.  They make printers, for crying out loud.  Nobody wants a Lexmark MP3 player either.
• Not organized around a standard platform.  MP3 is the defacto standard for ripped music, but not for online music (where Apple leads).  As music sales continue to migrate to online channels, MP3 will lose relevance.  (Unless the labels decide to go DRM-free.)

Given this misalignment, I'm not sure why Dell decided to get into this market in the first place.  It's clear that the winners in this space will continue to look like the current strong competitor: innovators with strong leverage over the online sales channel.  Product innovation isn't really a part of Dell's DNA, and they exerted zero leverage over the Windows Media channel.  This was destined to fail, and it shows maturity that Dell is dropping out before blowing even more money chasing Apple.

California Server Power Initiative

eWeek is running an article describing a rebate program that California's PG&E has instituted to reward companies for using more energy-efficient servers.  As I have written before, server energy consumption is rapidly becoming the new battleground in the hosting arena.  This rebate really puts some financial weight behind the argument for eco-friendly servers.  Whereas the servers may cost a little more up front to buy, the savings in electricity costs combined with the rebate, make the argument less cut-and-dried. 

This is obviously a brilliant marketing move by Sun.  If it's not matched by other states, this will also add gravity to California as the best place to host servers (much as the South is a magnet for auto manufacturing plants).

Web 2.0 Valuations In Perspective

Just in time for Friday, the 9rules bloggers have posted an interesting take on the valuations of Web 2.0 companies.

Also, Paul Graham writes his take on the Kiko affair.  In my initial post, I mentioned that starting with less capital can be an advantage.  Graham indicates that this is true even in failure.  By keeping expenditures low, Kiko's founders may be able to recoup all of their investment and move on to the next project without too much hassle.  So at the very least it looks like Web 2.0's inevitable failures will be a lot less painful than the Web 1.0 kind.  There's probably some fancy observation to be made about the economy's continued evolution toward greater efficiency, eh?

Kiko Liquidation

Kiko has put itself up for sale on eBay.  Kiko is an online calendar service that is notable because of its adoption of standards-based rich front-end technologies (AJAX) to build the GUI.  Or at least it was notable for that until Google launched Google Calendar.  Kiko clearly does more than Google Calendar, but Kiko's management clearly sees a trendline in Google's offering that they don't like.  No matter how much Web 2.0 buzz Kiko can generate (kudos to them for the level of mindshare they have achieved in such a short time), it pales next to the buzz Google can generate.  I think it's wise of the Kiko team to sell now, before Google revs their Calendar a few times.  Fortunately, Kiko is a YCombinator-backed startup and probably hasn't taken on a ton of capital.  So this exit might not recoup all of the capital that's gone into the company so far, but I'd but it gets back an appreciable portion.

Lessons learned:

1. an AJAX GUI is not a sufficient differentiator (i.e. to set Kiko apart from the years-old Yahoo! Calendar)
2. Minimal capital investments are a Good Thing

Good luck to the Kiko team in its next venture!

Google in Atlanta

Google continues to be a forward-thinker in its operations, most recently by opening an office in Atlanta.  Most big tech companies have offices around the world.  But for most, those offices are solely a way to be closer to their customers in various markets (which is a good thing).  Google is a large-scale leader in taking this a step further and using its far-flung offices to absorb the best talent available anywhere.  Engineers not working at HQ?  Unthinkable at many tech companies (except for the odd telecommuter).  But this is absolutely a great way to leveragemodern communications technology to help recruitment & retention.  After all, no matter how awesome you think your firm's HQ city is, some people will hate it.  The Bay isn't for everyone, and lots of people won't move to Seattle.  So Google brings the jobs to the people.  Definitely a winning strategy.

Tech Recession Likely

I've been talking to friends lately about what I think are the symptoms of a possible tech recession on the horizon.  I've been asked to put my thoughts together, so to help me organize my thoughts I'm going to post a draft here.  With any luck, I'll publish the final product here if/when it happens. 

I think we're due for a recession in tech for a number of factors, but the driving factor is that I think we've been a bit weak on achievement-focused innovations this cycle.  By achievement-focused, I mean innovations that allow people to accomplish things they couldn't do before.  Email let people keep in touch in near real-time; desktop publishing let individuals and small businesses do their own printing; the Web lets people do business in places they couldn't before the Web; GUIs let novices adopt computers to do all of the above.  All of these are achievement-focused innovations.

Contrast these with innovations that allow people to do things better or cheaper or easier than they can now.  Examples of these technologies include faster chips, graphics cards, online faxing, and wireless Internet.  This isn't to knock these innovations; we need faster/better/cheaper too.

The difference between these types of technological innovations is important in determining the health of the technology purchasing cycle, because it helps explain the motives behind purchases.  Achievement-focused technologies get purchased seemingly out of proportion to their immediate environment, because there is no other way to get the job done.  Cheaper/Better gets bought when the old way of doing things isn't good enough anymore.  Obviously there is some overlap but this is the basic idea.

So where are we now?  In the last 4-5 years, there has been a ton of innovation, but all of the major products seem to be in the cheaper/faster camp.  Examples by big-name company: Cisco: VOIP & wireless Internet; Microsoft: refreshes of Office and (eventually) Windows; Intel & AMD: multicore (i.e. faster) chips; Yahoo!: tons of incremental improvements to its service; Google: a better spin on a decade-old industry; Salesforce.com: Siebel over the Web.  (Contrast with Apple, the leader in the counterexample spaces of producing video content at home and managing digital media.)  So this cycle has largely been about consolidating our gains from the last boom and improving on those technologies.  But history tells us that this cycle will only last another 2-3 years, so we are at best most of the way through, and we have yet to see the go-go type of spending we'd expect from a tech boom.  A lot of people are taking this to mean that tech is mature, but I think that's not quite correct.  I think what we're seeing is the natural product of a weak macro expansion combined with a Better/Cheaper phase in technology innovation.  (After all, how mature can an industry be when startups like Salesforce.com can still throw turmoil at established firms?)

If that's where we are now, what can we expect?  My expectation is that this macro expansion will be shorter than the prior expansion, and then we will see a recession of some magnitude.  But after that I'd expect that the next technology up cycle will include many more achievement-focused innovations, which will make it look like 1995 again ('99 will still be an outlier in growth from the prior year).

Stay tuned for Part 2, where I outline my reasons for thinking the recession in tech is closer than many believe.

Fastest Internet in the U.S.

Verizon has raised the stakes in the broadband access game by raising its top offering to a whopping 50Mbps.  With Cablevision's OptimumOnline topping out at 30Mbps, this puts Verizon squarely in the lead.  Stupidly, Time Warner tops out at only 10% the speed of the high-end Verizon offering with no option to pay more for faster service.  With DSL beating Time Warner's speeds over the superior cable medium, one has to wonder how much of a priority broadband is at Time Warner.

Electricity as Strategic Advantage?

The Wall Street Journal is running an article today about how Internet firms are increasingly locating their facilities with electricity cost as a prime consideration ("Surge in Energy Use, Energy Costs Has Big Tech Firms Seeking Power" -- $$ Registration Required).  It's no surprise that the rising cost of energy is affecting firms that run tens and hundreds of thousands of machines.  What I found interesting is just how big this demand has gotten: the article mentions that Austin Energy attributes 8.5% of its total demand to data centers.  This, from an industry that was essentially nonexistent 10 years ago.  As the physical footprint of the Internet gets bigger, the incentive to fundamentally improve the power consumption of the machines will only grow.  Perhaps Sun is onto something with their "eco-friendly" campaign to increase power/watt using their CoolThreads technology.

Financial Websites Compared

For years, Yahoo! Finance has been my favorite site for basic financial data.  I've used most of the other big sites, but I've found that Yahoo is consistently the best for my needs.  Specifically, they got the major things correct:

• provide a ton of data
• in a concise format
• as a quick download so I can pull it up quickly
• and without making me use it in tandem with another site to find out a particular piece of standard data (i.e. shares outstanding)

Enter Google Finance, Google's entry into the financial information space.  Google takes a slightly different tact than Yahoo! and uses a rich Flash control for the stock chart.  This allows the chart to be customizable on-the-fly.  Subjectively, it also feels like changing the chart's time period is quicker than with Yahoo!  (I should also note that the time periods can be customized in ways that Yahoo!'s can't.)  Also, Google scrolls the news in an attempt to be contemporaneous with the displayed chart, so that you can overlay historical news against historical stock trends.  It looks like the news data doesn't go back beyond about 18 months, but I'm willing to give Google the benefit of the doubt that they will be able to index and add more historical news stories.

So generally I'd give the nod to Google.  But they fail miserably in one key area: there is no dividend information.  So for now I use Google Finance as my launchpad and switch to Yahoo! to get yield data.  I'm hoping Google fixes this, or it's back to Yahoo! I go.