Web 2.0: Shifting from “Get Fast” to “Get Massive”

What we have Learned from the First DotCom Bubble

People that were in the business in 2000 like to speak of the first dotcom bubble. This is a period when we all learned a lot about how to do business on the web with new technologies.

At this time, the main challenge and the ultimate goal was to be first to reach a market. The first dotcom era was about developing faster than the competitors. It was a stressfull and crazy period: We were all closely watching the competition to compare anxiously our progress rates. It was like the end of the world if competitors were doing an announcement a few days before ours.

The explosion of the dotcom bubble reinforced this strange feeling. The companies that had made money from the dotcom period were the ones who managed to sell early. The company that ha reached a niche market first had been the most successfull.

We have to rethink everything in the Web 2.0 era

Two years ago started the second era in the "new economy". Investors were interested again in web startup companies. Web sites and services were becoming valuable again. Businessmen and development teams that lived the first bubble thought that the same key success factors were going to be the same again. "Develop fast" was their motto. That's why Ruby and Ruby on Rail became increasingly popular. The people at 37Signals made it popular by showing to the world that you could develop a new service for the web from scratch very quickly. The Ruby on Rail video had shown how to get things done fast.

Get Massive

However, the key factors to succeed in the Web 2.0 are quite different this time. The first bubble had forgotten one small detail. To make a living of free services on the web, you have to get massive. First, you need a lot of users coming to your site. Then you need an infrastructure that won't crash when you get a lot of users. Finally you have to work out how to turn this interest into something that earns money, so you can keep the site going.

This is the heart of the Web 2.0. That's what people are calling the "social web". From the user point of view, it is about social collaboration to produce contents, like demonstrated by Wikipedia, YouTube, and many others. From the company point of view it is about getting massive.

Getting massive has many consequences in every part of the company. This means that you need an adequate marketing to attract users and that you need a trully good idea and good product. It is not about being fast but about being able to get users and to keep them. This trend is also deciding in technology choice. It means that you need a technology that can scale to amazing level with a reasonable effort. Companies do not need an enabler for fast development. They need a development environment that allows them to reach the mass with a limited amount of development effort and a reasonably sized hardware platform.

Erlang becomes the key enabler

This is where Erlang enters into play. Erlang is a concurrent language which is users to develop naturally clustered large scale applications. This is one of the only languages today that has been designed to get massive.

Tim O'reilly has published a blog post that shows this trend: Concurrent Programming: Erlang, Haskell...and XSLT. He compares the trend of Ruby and Erlang and this shows that Ruby is not progressing anymore: Ruby website visits are decreasing. On the other hand, there is a rise in visits on Erlang.org webiste. The figures are extracted from Alexa, but from what I can tell the figures are confirmed directly from Erlang.org statistics.

If I am not mistaken, this is only the beginning of the Erlang era. It is very likely that Joe Armstrong new book will play a big part in helping spreading Erlang in the coming months.

Web 2.0 = Get Massive

This is also a trend we can confirm at ProcessOne. Companies are contacting us because they need help to get massive in various area. They want a massive Instant Messaging infrastructure because the social element makes Instant Messaging a needed feature for every Web 2.0 business. We help them by deploying and integrating large scale XMPP-based Instant Messaging server on their domain. They want web sites that can sustain a massive frequentation and we help them facing the load with Erlang technologies, such as Yaws (See the Apache vs Yaws paper). They want massive scale infrastructure and we help them develop the needed pieces of software (load balancer, clustering layer, etc), with Erlang, the massive glue language. They want to check how many users they can sustain and we help them with our Erlang benchmark tool.

In this domain, ProcessOne rules and all we can say now is that we power the massive web :)

In the "massive web" war coming, we are one of the most active ammunition providers. We have exciting times ahead.

Comments

A competitive advantage of Erlang software, is that Erlang is suited to solve today’s energy and heat issues in data centers when it is combined with multi-core processors such as Intel’s Teraflop Research Chip: http://www.eetasia.com/ART_8800453831_1034362_db7dce9c200702.HTM

PS@Mickaël: Can you try to contact Intel to ask if they would allow you to do a benchmark of ejabberd on this chip? B-) Time to break the 10 million concurrent users per node barrier! Or maybe even more ;-)

Posted by sander on 05 Mar 2007 at 16:18

Erlang is really the next best thing!!. Really folks it is a great enviroment and programming language. It is Dynamic, Functional, Modern, Concurrent, Distributed, Parallel, RAD and Fast on multicores that is also the next best thing on hardware.

Posted by alpha on 27 Apr 2007 at 07:17

Erlang becomes the key enabler,I agree a lot :-)

Posted by Haiming on 11 Apr 2008 at 06:15

I agree.  Erlang technologies is “the” enabler on a massive scale.

Posted by Jim Bisnett on 15 May 2008 at 18:50

I agree with you that Erlang is a prime technology for the next wave of startups. What I disagree with is the why.

Sure, getting massive is important, and yes erlang will help you scale out. But I think the reason that erlang is important is because of what users expect.

Users are increasingly expecting to be able to interact with each other through the web. To be able to push, prod, and poke each other. And this requires webpages that aren’t just fancied up database extracts. It requires websites to be hosted on communication hubs.

And for that need, what better than a language designed to run communication hubs?

Posted by Brett Morgan on 20 May 2008 at 13:12

Add comment