Wow, Google Dev Day 2010 in Prague is over and there was lots of interesting information presented in the sessions. Below some notes I took during some of the sessions.

Native Client

• Currently supported languages: C, C++. There are plans to extend the platform for other languages like C# and also to integrate with IDEs like Visual Studio and Eclipse.
• Currently works reliably only in Google Chrome; can be used for extension development.
• Possible use cases: client side encryption, image processing, games
• Native Client is sandboxed in two layers: Native Client Sandbox, Google Chrome Sandbox.
• Additionally certain IO and system calls are blocked.
• Code must be verifiable, which has a size and performance impact.
• http://code.google.com/p/nativeclient/

In my opinion Native Client probably won’t become popular in main stream applications, but I see a lot potential for Chrome Extensions.

What’s new in Google App Engine

• AE for Business has been released recently. Main highlights are a Service Level Agreement and SQL support.
• New features and API’s
  • Channel API: bidirectional client server communication.
  • Mapper API: Support for full MapReduce. Ideal for dealing with large datasets and distributable problems. App Engine Map Reduce.
  • Multi-Tenancy: Very easily usable by implementing a servlet filter and setting a namespace.
  • Matcher API: Document matching infrastructure.
  • High Performance Image Processing

All in all App Engine made a much more mature impression than last year. Together with SpringRoo, GWT and the Spring Source Tool Suite it’s a great environment for rapid application development …. if you are into Java …

Storage, Big Query, Prediction API

• While App Engine is a Platform As A Service (PAAS) offering, Storage, BigQuery and Prediction API are the first products of Google’s Infrastructure As A Service (IAAS) offering.
• Storage: essentially like Amazon S3. Compatible with S3′s REST API.
• BigQuery and Prediction API builds on top of Storage
• Data is stored in US data centers, but distributed through Google’s global network. Thus latency is very low.
• Currently no Amazon-like areas are supported.
• Comes with a command line util: gsutil. gsutil can be used together with S3, too.
• Prediction API is an API to Google’s machine learning algorithms. Essentially you can upload a dataset to Google Storage, train the algorithm and then ask for a prediction based on a new input. The algorithms themselves are a blackbox; you don’t have any influence and no way to configure / fine tune them.
• BigQuery: lets you query large datasets with a subset of SQL very quickly. Great for reporting.

Android News

New in Android 2.2

• Licensing Verfication Library and Service
• Application Error Reports (integrated into the market place, including the possibility for customers to provide feedback)
• Just in time compiler: makes native apps 2x – 5x faster
• Uses now V8 Javascript engine: makes web applications 2x – 3x faster
• Device admin API: Enforcing enterprise Security Policies (e.g. max password length, etc)
• Cloud to device messaging: Great for server push. Saves battery on the client side and provides a much better user experience.
• App Data Backup API: Additionally to restoring the users apps on a new device, this gives the developer the opportunity to restore the users application data, too
• Speech API
• Easier configuration of “Install on SD”: simply configure this in your manifest file.

The article has three parts:

Part 1

• What is cloud computing
• Driving the hybrid model (outsourcing of specific services to the cloud)
• Amazon Cloud Services
• Google AppEngine
• Microsoft Azure
• Salesforce.com

Part 2

Implementing a hybrid application: reading messages from local queue, sending to Amazon SQS and reading it again back from there

Part 3

Security considerations

• Cloud Governance: Define _local_ policies for how to use the cloud. Usage of XML Gateway is of advantage, e.g. Vordel XML Gateway
• XML Gateway can be used additionally to encrypt data before it’s send to the cloud or to strip out private information.
• How does Security work for Amazon SQS and S3
• Amazon SQS Policies

Enjoy reading!

Amazon today announced a new product called Virtual Private Cloud. For now it’s available as a limited public beta, but still it’s amazing news because it will make the cloud concept more enterprise-friendly and can be set as an argument against the typical security concerns corporations typically raise against shared environments.

For further reading check out TechCrunch’s article: Amazon Eyes Big Enterprise Budgets With Virtual Private Cloud Service

2009 is the year of the cloud. Google started to support Java on its AppEngine platform, Microsoft released the Azure platform to run .NET applications in the cloud and if you’re running a Rails app you might want to take a look at EngineYards cloud offererings or Heroku.
Additionally to these PAAS (platform as a service) kind of clouds you find more open / configurable platforms: Amazons Elastic Cloud or Joyent are two examples. There are companies building up on these platforms like RightScale and Cohesive FT. Both are operating on EC2 and helping you manage the cloud.

Cloud services are a great way for startups to deploy their applications on a scalable platform without the need to invest thousands and eventually millions of $ to build up a reliable data center to deliver the service they have in mind. They don’t need to hire a whole staff to take care of their operations and they don’t need to sign expensive support contracts for hard- and software components.

One often mentioned concern about the cloud approach is security. There are applications which are critical: financial applications, applications in the health care space to mention two examples. One solution for this is the “private” cloud, meaning basically a private data center …

Next to security there is another point you should consider. I haven’t really thought about that by myself, not until I read this fantastic article on Forbes: Analyzing The Cloud.
In short: Who do you expect will your customers be? If you sell your product mainly to private customer or small and medium sized companies you’re probably fine. Bigger corporations tend to have already big data centers and complete staffed operation departments and it’s less likely that these want to move parts of their operation outside the company.

According to this article Google says there are “over 1.75 million businesses, schools and organizations” using Google Apps.
This is far more than I thought. No wonder that Microsoft announced the Microsoft Office Web Apps as part of Office 2010. Everybody wants a piece of the cake!

No matter which IT news channel you read, there’s no way you missed the dozens of articles about Cloud Computing which have been released in 2009 up to now. It looks like 2009 is the year of the Cloud, more than ever. In fact I believe that the Cloud is a great model for many applications, eventually even more than you’d think right now.
There are some things though you want to consider when planning to move your app to the Cloud. These two articles are good starting points:

• Lost in the Cloud by Jonathan Zittrain
• Cloud Computing by Bruce Schneier

There was an article on TechCrunch about what can happen if all the data of you company is in the Cloud: The Anatomy Of The Twitter Attack.

All the risks which come along with the Cloud Computing model don’t speak against the model itself. It’s all just about mastering these risks to make that model successful long term.