SmoothSpan Blog

For Executives, Entrepreneurs, and other Digerati who need to know about SaaS and Web 2.0.

Archive for November 16th, 2007

Cloud Computing in Someone Else’s Cloud: The Future

Posted by Bob Warfield on November 16, 2007

Ever hear of a fabless chip company?  This is a company that sells Integrated Circuits but owns no manufacturing facilities.  They just write software, in effect, and send it out to someone else’s fab.  Brilliant.  Many kinds of manufacturers often do the same.  After all, manufacturing may not be the distinctive competency of a company, or the company may achieve better economies of scale by using centralized manufacturing owned by much large companies.

This is starting to happen big time with web software.  IBM just announced they’re going to join Amazon in the cloud computing business with “Blue Cloud”.  Companies will be able to buy capacity in someone else’s cloud which they sell as their own.  No need to own any hardware or even visit a colo center.  Why would you want to own a datacenter if you didn’t have to?  Why would you think you can do it as well as Amazon or IBM?  Many others including Yahoo, Google, and Microsoft will be a part of this future.  Sun is already there with Sun Grid. 

So far, the formulas are pretty similar.  IBM and Amazon are both Linux-based systems built on virtualization software.  At some point, if enough hardware capacity is locked up in this rental data centers, it will become an important sales channel for all server hardware manufacturers.  Take Dell for example.  They’ve always sold direct.  Shouldn’t they consider this kind of business, especially when other hardware companies are going there?  What about HP?  Look at it as a way for hardware makers to switch from the equivalent of perpetual licensing to the SaaS rental model. 

What about Microsoft?  Can .NET be as successful if they don’t build a Cloud Computing Service that is .NET based?  Seems to me this is a strategic imperative for the OS crowd lest Linux steal the show.  Sun is already there with Solaris on Sun Grid.  This is the system my old alma mater Callidus Software uses to host their SaaS solution and it works well.  IBM is not missing the chance to offer PowerPC as well as x86 servers for Blue Cloud.  IBM is also partnering with Google around Cloud Computing, so there may be all sorts of interesting bedfellows before this new paradigm is done rolling out.

A great example that’s being written about by Scoble and others is Mogulus.  CEO Max Haot says they don’t own a single server, it’s all being done on Amazon, and yet they’re serving live video channels to 15,000 people with just over $1M in funding.  You’ve got to love it!  A number of other serverless and near serverless companies commented on Scoble’s post if you want to see more.  These big guys are not the only ones in the business.  Certainly companies like OpSource and Rackspace count too. 

There are many potential advantages, and a few pitfalls.  First the advantages: it’s a whole lot easier and cheaper to build out your infrastructure this way.  Why have anything to do with touching or owning any real hardware?  How does that add value to your business?  The real innovators will make it easy to flex your capacity and add more servers on extremely short notice.  Take a look at your average graph of web activity:

CNN Traffic

This is traffic for  Notice how spikey it is?  Those are some big spikes.  If you web service hits one, you must either have a ton of extra servers on tap, or deal with your site getting painfully slow or going down altogether.  With a utility computing or grid service such as Amazon EC2, you can provision new servers on 10 minutes notice, use them until the load goes away, and then quit paying for them.  Payment is in 1 hour increments. 

I know a SaaS vendor whose load doubles predictably one week out of every month because of what his app does.  He owns twice the servers to handle this peak.  He’s growing fast enough at the moment that he doesn’t sweat it much, but at some point, he could really benefit by flexing capacity.

Now let’s talk about downsides.  First, most software doesn’t just run unchanged on these utility grids.  Even if it did, most software isn’t written to dynamically vary it’s use of servers.  Adding servers requires some manual rejiggering.  Amazon has a particularly difficult pitfall: you have to write your software to deal with a server going down without warning and losing all it’s data.  In fairness, you should have written your software to handle that anyway because it could happen that you whole machine is toast, but most companies don’t start out writing software that way.  There are companies, Elastra is one, that purport to have solutions to these problems.  Elastra has a MySQL solution that uses Amazon’s fabulously bulletproof S3 as it’s file system.

The second issue isn’t so much a downside really.  We can’t blame these services for it at any rate.  What I’m talking about is automation.  To really take advantage here you need to radically increase your automation levels.  I recently saw a demo of some new 3Tera capabilities that I’ll be writing about that help a lot here.

The bottom line?  You’re missing out if you’re not exploring utility computing: it can save you a bundle and make life a lot easier.  The subtext is that there are also a lot of new technologies, vendors, and partnerships coming down the pipe to help maximize the benefits.

Related Articles

Nick Carr picks up the theme.  One of the commenters raises an excellent point.  Using an IBM or Amazon gives peace of mind to customers of small startups.

Posted in business, data center, ec2, grid, saas, strategy, Web 2.0 | 3 Comments »

Universities Should Podcast Every Class

Posted by Bob Warfield on November 16, 2007

I started out a bit skeptical, preferring to read than watch, but I’ve gotten to where I like short videos as a way to capture experiences quickly and easily.  I appreciate the difficulties of interviewing all the more as I’ve done several and struggled to catch everything and still focus on driving the interview.  As I was reading Scoble’s post on ClearTXT the thought popped into my head:  Universities should record and podcast every lecture.  Why not?  They don’t have to release them outside if they’re concerned about protecting their IP in some way.  It should be completely unobtrusive to equip the lecture halls to capture video of every class.  A relatively simple and inexpensive system could tie into the class schedule, capture video with sound directed at the where the lecturer stands, and put all of that onto a web server for students to access.

Will it reduce attendance in class?  Perhaps.  Teachers can still use the time honored tools ranging from taking attendance to pop quizzes to critical handouts that don’t go up online. I can’t tell you how many times in my own past it would have been useful to refer back to a lecture, perhaps as I was studying for the final.

Schools that take the next step and “Open Source” their lectures would really be doing a tremendous service for all.

Posted in Web 2.0 | Leave a Comment »

Oracle Fusion Will Come In With A Whimper

Posted by Bob Warfield on November 16, 2007

I just read over on ZDNet that Oracle’s initial Fusion offerings will not be real applications at all.  Rather, they’re CRM applets.  Apparently Larry Ellison really floundered with questions about where the real suite would come from.

It gets worse.  When I saw the screenshots, I realized these are the modules Oracle is building Flex UI’s for.  What’s wrong with that?  Nothing, I love the Adobe Flex product, I’m using it myself to build the SmoothSpan client.  But, and this is a big but, Flex is an extremely powerful tool for producing Flashy UI.  It makes me wonder whether Oracle has invested at all in the backend behind that UI.  When I hear they’re building minor modules, not real applications, and the minor modules are Flex, I start to have visions of very very few people actually working on Fusion at all, or at least very few working on anything that’s coming out near term.

That’s in huge contrast to SAP’s boil-the-ocean By Design product, but it’s completely in keeping with what I’ve heard from Oracle people I’ve talked to.  For them, Fusion was something someone else, someone they didn’t even know was working on.

I see this latest news as further confirmation of what I surmised when John Wookey left:  Fusion is badly off track.  We may or may not ever see a real rewrite of the ERP suite to produce a modern architecture.  What is certain is we won’t see it any time soon.

What remains is to ponder what impact this will have on the company.  They’re doing extremely well, so it may not matter.  What pressure they’re feeling comes from SAP, SaaS, and Open Source.  SAP isn’t exactly racing By Design to market for their own reasons.  SaaS and Open Source are much bigger threats.  If the world finally figures out that MySQL can be made to scale as well as Oracle as many of the biggest web businesses have proven, it will be a dark day for Oracle because the DB franchise is the heart and soul of that company.

Oracle has tried some counterpunches.  They went after RedHat, but that doesn’t appear to have produced results.  They say they’re going after the SaaS CRM market for big enterprise.  We will see how much impact that has.  Currently, their own SaaS business is large by most standards at over half a billion dollars, but very unprofitable according to their last 10K. 

Still Oracle can keep doing what it has learned to do best: it can keep acquiring.  At least for a while.  At some point, it will hard to find enough fish of sufficient size to gobble up that will make a difference.

Posted in business, strategy | 1 Comment »