I’ve been exchanging emails with Dale Vile of Freeform Dynamics on the subject of the cloud and PC applications. I’ve abstracted some of the points that were in the email exchange, because I think they raise important issues. Let’s draw a few lines in the sand in respect of PC and PC cloud apps:

• A massively scalable architecture pointed at a single transaction type or a small number of transaction types provides the cheapest per user costs in the world, not by a small, but by a large margin. That’s why corporate Gmail costs a mere fraction of a Microsoft Exchange implementation. Massive scalability cuts the per-user costs to ribbons, when you have sufficient user numbers.
• Right now, Google Apps are often used in conjunction with MS Office and/or Open Office. There are several reasons for this, but the main one is that RIA hasn’t taken over yet. As regards Google Apps, Google is developing code so that all its apps can run locally when there’s no connection. Google Docs, Gmail and Google Calendar already have local versions – although they’re in the first release. That by the way, is a reasonable definition of RIA.
• The churn rate on Google Apps is very low. It is estimated to be about 0.5%. Clearly, companies really are migrating to Google Apps, an idea reinforced by the fact (according to Google) that half its customers for Google Apps are paying customers. Where dual app usage exists, corporations will not want to proceed with dual app usage indefinitely.
• As of Jan 2009, Google Apps has about 1 million business customers (companies) and 10 million users. The rate of growth is currently about 100%. Incidentally, Zoho has 1 million users and similar growth rates.
• Massively scalable architectures can deliver response times in the 400 millisecond area, which is often as good as or better than a local PC app and nearly always as good as a PC app when the interface work is done locally (i.e. RIA).
• Nowadays very few Mac/PC users have any idea where any program is executing. In most instances there are two scenarios:
  1. An application is purely local, although it will normally be maintained from the cloud by download.
  2. An application is dependent on a web site, in which case some code executes on the Mac/PC directly (usually Javascript), some code is executed by the browser (which means HTML and CSS), some code executes on the server (PHP, Python and Ruby).

I’ve monitored cpu resource usage on the Mac and by far the biggest user of resource is Firefox (or Safari) – unless I start messing with movies or photos. Firefox uses more cpu than all other apps put together. It uses lots of memory too. It’s not the browser per se it’s what the web sites throw at the browser. If they insist on being in constant contact, like TwitterFall for example, it chews up cycles for communications.

A random survey of what I’m running at this very moment indicates 10 apps out of 27 that have an internet component (Mail, Skype, Devonthink, Forklift, Forefox, Safari, NetNewswire, Coda, iTunes, Parallels). None of these except Mail are traditional PC apps and 3 are for web site development. But if you count open web pages, 7 from 11 tabs open in either Firefox or Safari are applications (Twitter, WordPress, Google Search, Ezine Articles, Talking Alpha, etc.) rather than static web pages. A good deal of what I’m doing at any time uses the Internet.

What happening is that the web is gradually infiltrating my computer usage. So let’s have a diagram. It may not help, but I like drawing diagrams (next page):

« 1 2 or View All »