South Pacific Web Mapping

With the plan of introducing a new South Pacific fibre (see: http://www.pacificfibre.net)

the New Zealand Internet Industry will get a major boost in its potential growth.

The existing Southern Cross cable architecture introduces a significant latency factor due to the commercial decision to terminate all legs at Maui (Hawaii).

This doesn’t have a major effect on data speed and bandwidth (with powerful routers in Hawaii) although introduces major challenges for VoIP communication where the call quality is determined by latency (number of hops and carrier distance).

With a direct link from California to Auckland and a straight cross over the Tasman either to Sydney or Melbourne (see below):

(source: http://www.pacificfibre.net)

and  the rapid growth in VoIP usage across the pacific, Australian VoIP providers and voice operators such as CRM and call centres will find NZ  a very attractive environment, located close enough to Australia and right on one of the Pacific’s best communication hubs.

We are currently experimenting in SIP packets routing thorough Asia(Singapore) and US (Virginia) for optimizing global voice link to Europe. stats soon.

Thanks to Vladimir Verlinsky for the TCP/IP BGP4 routing leads.

Last year while chopping some wood down the garage I found this old newspaper (the other side of this page was about Ronald Reagan campaign for the 1981 US elections, not sure how this one managed to stay there all these years…

my Coromandel mates told me this summer its up on the agenda again (see: Coromandel WatchDog).

We lived in the Coromandel for over two years and had great two weeks in this beautiful green bush on Xmas,  will be back as every summer next year!!

a cool Coromandel itinerary I have recommended for several of my overseas mates could be seen here.

Had a bit of experimenting on multihoming using BGP, the main idea is using BGP4 to announce an address space to all upstream links out side the autonomous system (AS) . This is particularly good for scenarios where a routing path requires high availability (allowing swapping of a cluster while serving web requests from different data centres with no down time), but is also experimented for load balancing global resources (serving the same web content from different locations).

I have got into BGP4 from my work on GeoWeb apps a few years ago when I realised that with BGP4 and dynamically created routing tables you cannot rely on IP address alone for identifying a requesting connection’s physical location (see here).

The bottom line is that when coming to implement a fast loading rich mapping solution(or any other mash-up), it is significantly important to know the client’s physical location (for serving content from closer locations) and  knowing its IP is not enough any more!!

possible scenario from a New Zealand Australia optimisation case(implemented within a Melbourne mash-up)

more on this

The known relation between slow pages and revenue loss is finally in numbers, a joint announcement of search engineers from Google and Microsoft’s Bing has released some stats on a long term research done proving that users will spend less time (and money) while navigating accross slow loading sites (see bellow):

The full article from Oreilly.com here

This is another reason to deal with latency and heavy pages for maximising profit within e-commerce sites on top of saving on bandwidth and traffic costs.