The Themes of Structure ‘09

Primitives
The two biggest web technology players at the event (Facebook, Google) used this term often in referring to their programming discipline. The Register has a nice piece covering a passionate exchange between Microsoft and Google engineers on adherence to consistent primitives. Google said MS would fail at matching its speed because they lacked discipline around programming simplification. Google forces developers into narrow development frameworks driven by GFS, map-reduce and big table. If you want to create a service it must be built on those–end of story, game over.

Facebook then revealed much of the original PHP-MySQL code originally written by Mark Zuckerberg is still in use today. (Mark is an ace coder though, who was recruited with big money while still in high school by MS and AOL) They also emphasized a very disciplined approach to keeping simple primitives as the core. This has allowed them to scale primarily by buying lots of lots of servers without ever having to change much of their core code.

Microsoft’s Azure product manager made Twitter #structure09 waves when he suggested very few people know how to scale the LAMP stack. Azure, he promised, would solve this scalability problem as a feature of the platform.

Overall code primitive design was foremost in the mind of the two web leaders, far surpassing infrastructure optimization, or even application features. Primitives are the new infrastructure for world scale web aps; servers/data-centers are now referred to as ‘atoms’

“Data centers are just atoms. Any idiot can build atoms together and then create this vast infrastructure”. —Vijay Gill, Google

Its a trend not a technology:
The conference speakers each offered quips about the definition of clouds. While everyone’s explanations were taken seriously, the biggest reaction was one of relieved laughter at Russ Daniels’ quip “Why are we so excited about the cloud?” Daniels asked, “One answer is everyone can draw a cloud.” It was clear despite the a conference being organized around the term, some healthy debate on nuance remained.

Ultimately speakers agreed on one thing above all, cloud computing is a macro trend not a technology. Trends aren’t easily bottled in a concise definition; they are multifaceted and manifold by nature–they spawn different instantiations, and propagate along a vector of continuous change. So what exactly is that change? I’d point above all to a new set of programming primitives where parallelism and massive scale are built in. At some point these new primitives become the new ‘infrastructure’ of IT. Google has this one dead right.

Azure is exactly the correct strategic play for Microsoft. We will see how the execution of the strategy goes, but they are going after this new development model, where cheap ‘atoms’ come built into the IDE.

The systems providers will be fun to watch as this unfolds; this trend is all about reducing their ability to differentiate on an application by application basis.

Legacy vs. Cloud
Legacy vs. the cloud was discussed by every speaker connected to a legacy (so not WordPress, Joyent, Google, or Facebook). While many journalists and some analysts say things like “will X business application move to the cloud,” this generally has a simple answer “No.” Applications do not like to move, and never have. Why does HP still have a Tandem business? Enough said.

Secondly there have not been many highly stateful, transaction based cloud apps. It will come, but for now many process based IT apps will stay on traditional infrastructure.

Having been an enterprise IT salesperson and marketer, what’s most disruptive about cloud computing is that it consumes the marketing oxygen of a sales pitch unless its included. So expect cloud to be in most sales pitches for legacy infrastructure, somehow, someway. This in turn gives rise to lots of noise in the press–and you can see the challenge. Keep your eye on changing development models and you’ll be at the center of the change instead of on the noisy marketing driven periphery.

Marc Marc Benioff & Werner Vogels Swagger:
As mentioned above sales and marketing oxygen are all about trend alignment. No one had more confidence and swagger than Marc and Werner. Marc was hilarious in making fun of other industry players “Is Azune out yet from MS?” and Werner grinned from ear to ear saying “This isn’t the future, this is happening right now. Right now someone in your enterprise is on AWS buying services.”

These two companies have the cloud trend wind in their sails and it will only get stronger and stronger. It was a turning point from last year when it was more about the future than the present.

Just say no to servers:
Servers were essentially mentioned twice the whole day, other than in passing as a component of the cloud. “The biggest mistake we ever made was buying servers.” Matt Mullenweg, Founder, WordPress, followed in the afternoon by Facebook’s cajoling of sytems OEMs.

Storage:
The #1 problem described by almost every panelist operating a cloud infrastructure was storage. Its currently poorly virtualized and hasn’t gotten faster in years. This forces them to do unnatural acts of engineering and spending to grow storage and keep it cheap and fast. Everyone was excited about the coming wave of flash based storage devices–they are hungry for fast cheap storage. Does a hot problem leave the door open for a hot new set of companies?