Amazon VPC: Chasing the Enterprise Market

One of the persistent criticisms of Amazon Web Services is that it does not address the needs of enterprise customers. The most commonly cited shortcoming is its lack of network control -- the ability to segment the network, place firewalls between segments, etc. As an overarching criticism, the fact that one organization's resources share computing resources with every other organization -- be they upstanding enterprises or smarmy fly-by-night scamsters -- puts many enterprises off.

No, load balancers are not inherently unscalable

My tweets about the SOASTA presentation by Dan Bartow last night at the SDForum Cloud Services SIG has generated a number of follow up tweets and LinkedIn comments. Trying to respond 140 characters at a time is difficult, so I thought I would do a short blog post to describe (my interpretation of) what Dan said about the common scalability breakage points.

I'm bemused by discussions I see that assert cloud computing's self-service means the end of IT. I saw a blog post today that said end users would write their own apps because of cloud computing self-service.

All of this overlooks a raw fact: writing applications is hard. Mashing up some widgets is definitely within the ken of end users, but creating a web-based, data-driven, scalable, elastic application integrated with the corporate identity system, incorporating monitoring, management, and security is not trivial. Or easy. Or simple.

Yesterday brought the news that Google Wave was being killed. Announced with great fanfare a year ago and really only available for a couple of months, its demise was announced by Google as a kind of "we try a lot of things and some of them don't work out ... and we move on."

I applaud Google's willingness to try new things and accept that some of them, inevitably, will be failures, but I am struck by the fact that Wave failed.