South Pacific Web Mapping

I have worked on this one lately:

click here for Australian / New Zealand Versions.

and starting to really like SharePoint 2007 (SP2 onwards), excellent .NET application container that allows you to define your cache policies and expiry periods(BLOB, Output and Object Caching) and fine tune expiry periods.

a Nice feature is when you really need to go spatial (GIS queries and mapping UI) you can use native SQL Server as your data store and still have content and media resources caching managed by IIS/Load-balancer.

When dealing with Australian and New Zealand mash-ups I believe the best way is maintaining a data store with well defined expiry periods (e.g. traffic vs. weather vs. external providers cached content) and later distributing its static elements into the cloud.

I have been experimenting with Amazons EC2 services for a real estate related research project (heat mapping commercial property stats) and Amazon has excellent .NET hooks allowing you to push content out to the cloud and later reference it within .net applications (or any other content blocks), this methods significantly reduce bandwidth costs/server maintenance overhead and improves response and serving time by reducing latency.

The architecture illustrated below will provide a SharePoint managed CDN for $0.15 USD per GB of traffic (based on Amazon’s current data transfer price plan):

This page has been constructed out of content distributed from three different continents, so the bottom line is:

“use out of the box content management tools and distribute its product using cloud services”.

I am currently within the last stage of defining my research thesis in computer science, this research will address methods for delivering GeoWeb applications while minimising the effect of network latency on user experience.

This proposal is a result of practical experience in designing and developing web applications for the New Zealand and Australian Internet market for the last 7 years, the research aims to provide scientific evidence and methodologies for minimising the effect of network latency on the load time of web content (see my post on this subject here).

With the latest advances in response/load-time monitoring and measurements services, it is now possible to collect statistics of monitoring agents across the globe and identify network trends towards optimising web delivery according to the originating content requirements.

A typical web based map mash-up could potentially include HTTP requests from several different sources (content providers, storage distributors etc.) spanning across several continents, selecting the optimal resources during the initial load of the page (the DOM Ready browser phase) will be the primary factor for efficiently serving web content.

Within my practical professional experience I have been maintaining and monitoring over 110 servers while optimising the actual web applications running on top of those servers for delivering rich graphic content. This optimisation included usage of CDN providers, requests source tracking and progressive download techniques while monitoring the results from numerous locations around the globe.

A notable network monitoring solution that I have been using within my research since 2006 is Pingdom, which I believe is the only response time monitoring solution that exposes an API. This feature enables querying its worldwide monitoring agents counters for generating real-time, applications specific, global response time reports and statistics.

In February 1882 the Dunedin sailed out of Port Chalmers loaded with New Zealand dairy products targeting London’s markets while overcoming the distance limitations and utilising the latest cooling technologies, becoming the first commercial refrigeration ship ever.

I believe that with the massive move towards SaaS technologies powered by the growing IT Outsourcing/Cloud Computing trend, New Zealand’s prime location and unique IP routing link could play a significant role in researching ways for overcoming network limitations for physically distributed services.

On the arrival of the Dunedin in London 98 days laterThe Times commented: “Today we have to record such a triumph over physical difficulties, as would have been incredible, even unimaginable, a very few days ago…” which sounds very relevant considering the development of the internet today.

I had the honor to complete a web mapping project for the KiwiExperience which is New Zealand’s major operator of bus passes for travelers and backpackers.

This mapping solution included mashing up operators information and available bus routes throughout New Zealand with spatial data provided using the Google Maps API.

click to view map

One of the main achievements in this project was to provide a rich fast user experience while distributing content over several continents for users across the world, this was done by following the latency requirements I have posted last year through my latency research ensuring the optimisation of South Pacific content for international customers.

With the latest take over of Sun Microsystems by Oracle and the global economic slow down, the ability of web content originated from AU/NZ data centers to provide competitive user experience world wide will highly depend on utilization of cloud computing and resources.

A good example will be solutions where content is stored and maintained locally but where possible combined with resources available by content provider positioned closer to the Internets backbone (Tier I providers neighbors) such as maps and media available by Google, Akamai, Flickr and Data stores such as amazon (AWS/EC2).

Being involved in several VOIP integration projects in its early days I have noticed the effect of latency on voice quality and the ability to minimise this effect by locating the VOIP termination servers (where the signal turns from IP to PSTN circuit i.e. traditional analog line) as close as possible to the major target audience.

Following the rule of “make the product at its source but serve it from as close as possible to its customer destination.” has been used by traditional supply chains for centuries and still applies to the latest web architectures for maximizing your clients user experience.

Update October 2009:

The site has also won a Microsoft NZ Collaboration Award ;o)