Web 2.0 with Justin Duewel -Zahniser

Microsoft Announces Azure Pricing

The cloud computing space just became a little bit busier. Microsoft has announced pricing for Azure. Customers looking specifically to run Windows instances in the cloud now have a price comparison point against the Amazon Web Services offering. Notably, Microsoft's hourly compute cost is $0.005 lower than Amazon's--half a penny. It is also $0.02 more expensive than Amazon's Linux cost. Microsoft does not, however, differentiate between standard and "high CPU" usage as Amazon does so that leaves open the question as to whether there's no upper price band transition envisioned by Microsoft, or whether that will change over time.

CPU time prices are lower on Google App Engine, of course, but that's a bit of an apples to oranges comparison, given that App Engine does not run an Operating System and allows the creation specifically of certain types of web applications. I won't bore you with a further play-by-play price breakdown, which I'm sure will be thoroughly provided by analysts and commentators, and in many cases already has. I do think it's worth noting that Azure is attempting to provide something quite different from AWS in one respect--a system ready made for running existing Microsoft applications as cloud services. Amazon offers more of a utility platform at its core with the ability to run applications as cloud services provided currently through more of a community model than as a core offering of the system provided directly by the Amazon team. I would suspect, then, that most of the direct competition will be over Windows applications and the applicability of Live Services to a potential project until such time as Azure branches out to Linux and non-.NET frameworks, if it follows that path and competes more broadly at the utility level.

One thing I didn't see from Microsoft, which hopefully will be coming, is a price calculator for Azure. Cloud computing has been heavily embraced and championed in the startup community for lowering the barrier to entry for software and SaaS offerings and cost is a very important factor there. However, as cloud computing heads more towards the mainstream with providers--like your own GXS--providing cloud-style computing environments for specific business problems or vertical industries, the simple utility price comparison becomes much less relevant than the use case fit. In any case, I'm thrilled to see more competition on the scene.

FULL DISCLOSURE: I have worked on AWS and Google App Engine as a developer and Azure as a partner.

On Google Chrome OS

A lot of buzz kicked up over the announcement of Google Chrome OS last night. In short, Google wants to build a a windowing system on top of Linux that is optimized for running web-based apps (read: Google Apps). The target is netbook devices, which further distinguishes this from Windows, Mac OS X and Linux distributions such as Ubuntu that provide an environment historically focused on local data storage and native applications.

Reactionaries are starting to predict the end of Microsoft. How many times have we heard that one? I don't see many web-based apps replacing the hardware-accelerated 3D gaming market, for example. Needless to say, this is interesting and it at least has the potential, alongside the evolution of modern smart phones, to change the face of IT. The target seems to be business users. A serious "instant boot," backup-free, ultra-portable operating system designed for email, chat, web browsing, document authoring, presentations and spread sheets for mobile professionals? Sound like you? Throw in some image editing (Photoshop Express anyone?) and you just described 90% of a lot of our working days.

Of course, there's also still that looming regulatory, security and privacy issue of business data stored in the cloud to figure out. But, we'll worry about that tomorrow. I'm going to keep an eye on this because I'm always interested to see how Trading Grid Online works in new browser offerings. A web-based, hosted, more native-feeling supply chain appliance is an attractive proposition.

GPS Cell Phones to Simulate Disease Spread in Japanese Elementary School

Geo-location technology in modern internet phones has interested me quite a bit of late. There's a market for location-aware gaming on the horizon, not to mention commercial and advertising opportunities for retail and travel industries. My wife pointed me to an AP story this morning about a Japanese experiment in which GPS on elementary school children's phones will be used to simulate the spread of contagious disease by tracking participant proximity. What a novel use of technology.

I imagine there are some privacy or safety concerns that could play into this, but the results of modeling disease spread using actual people in an non-controlled environment rather than a computer simulation could be quite useful. I'd like to find out if and how they plan to control for behavioral differences in the children under the assumption that they will be aware of the different implications of their contact with other kids during the study.

I'm resisting the urge to make some kind of joke about Logistics Visibility for school kids, here.

I enjoyed Lydia Leong's post about Scala at Twitter

I enjoyed Lydia Leong's post about Scala at Twitter. If you're interested in high-traffic web services and/or cloud-computing from a philosophical standpoint, the discussion of developer productivity versus infrastructure cost and related arguments is an interesting one to follow. I'd link you to the tweet from which I referenced Lydia's blog post, but I got yet another fail whale just now :-P 

My Own Little Cloud

Last Friday, my feed reader picked up a post from Adam Wiggins at Heroku on Personal Cloud Computing. He was describing some of what they do with Amazon's EC2 cloud computing platform for development work. They're hitting cost, growth and mobility issues with this approach.

Shortly afterwards, I ran in to EUCALYPTUS. It's an open source project that provides a Linux-based local emulation of the EC2 API. I came across this in the beta description for Ubuntu 9.0.4, which I track since my home computer runs Ubuntu Linux. Thanks to the order in which I encountered these bits of information, I was better equipped to understand why someone might use EUCALYPTUS. One might be to develop against EC2, but leave all the instance and usage costs to the production cycle rather than starting to incur them in development. Another might be the convenience of being able to work on this without regard to connectivity (in theory, although clustering tends to require a static location anyway).

This is a pretty interesting model, the home-brewed personal cloud. One thing utility cloud computing models have provided is a very low barrier to entry for technology startups. This helps the ecosystem flourish and gives formerly riskier businesses more of a fighting chance (though debate is still out on whether this mostly leads to widgets and ad impressions versus sustainable innovation). Imagine lowering that barrier even more by providing tools to build for that platform at zero utility cost prior to even rolling out the product. Very attractive.

SaaS developers. Go native! But, you know, not like Kurtz, of course.

Overcast

I've been appreciating Overcast, a good podcast on cloud computing. I highly recommend the episode on security and PCI compliance in the cloud.

James Lindenbaum on Instant Deployment

I caught an interesting post on the Heroku blog by James Lidnenbaum yesterday about instant deployment of applications.

2009 Tech Prediction - SaaS Will Finally Get to REST

In order to carry on the GXS tradition, I'm going to make a technology prediction for 2009.  Since I'm the guy in Marketing here who kisses his kids good night and then hacks away at code for fun, I'm going to focus not so much on the supply chain as on the world of Software-as-a-Service (SaaS).  You might want to add this article to your social bookmarking site of choice so you can check back at the end of the year and see if I deserve a coffee (or, in my case, a cup of tea).

In 2003, Amazon found that only 15% of their web service consumers actually used SOAP.  The other 85% used the REST URL-based web services interface (thereby prompting many people to ask, “what’s REST?”).  Since then, the Web 2.0 phenomenon and the rise of agile web development frameworks have increased the ease of use and availability of URL-based web services significantly.  The newest generation of web developers are living and breathing HTTP, AJAX, JSON and a host of other acronyms that prior generations may have considered fringe technologies.

SaaS providers will increasingly expose existing SOAP-based web services with an alternate REST implementation and build new services with REST in mind from the ground up in order to attract developers inside and outside their company to increase the value of their ecosystem.  Some of these developments will specifically target the growing market of internet smartphone users and the anticipated rise of netbooks and tablet PCs with development platforms that eat, sleep and breathe HTTP.  Additionally, startup technology companies who are developing on REST-based architectures and using these modern tools will continue to be acquired by established players, bringing them in to the SaaS ecosystem as well.

To quote Adam Wiggins, one of the founders of the tech startup Heroku, “Don’t fear the URL.”

Unit Testing Trading Partnerships

As a proponent of modern software development practices, I have a soft spot for unit testing.  It also helps that I spend most of my after work coding time in Ruby, whose community has unit testing in its DNA.  If you're not familiar with the practice, unit testing involves writing scenarios in code that prove a given object or method you have created behaves correctly.  

Sinatra: Micro-Frameworks and APIs

I've been playing with a Ruby micro-framework for web apps called Sinatra.  Billed as a DSL (Domain Specific Language) for minimal effort web apps, Sinatra actually seems to excel at a particular task near and dear to my heart:  URL-based web services.  If you're not familiar with a REST API, this is basically a web service queried entirely through structured URLs.  Something like http://twitter.com/statuses/public_timeline.format for the Twitter public timeline in RSS or XML or whatever.