Recently René, a friend of mine (and co-founder of the social networking website Metalcon), who is now researching in the context of Web technologies, has come up with a new concept how to efficiently retrieve Twitter-like newsfeeds in social networks, using graph databases such as Neo4j. If you are interested in Web technology in general or social networks in particular, I suggest you check out his article and the enclosed video, since it highlights how database technologies that deviate from the ubiquitious relational databases can be used and leveraged to produce a level of efficiency that is really stunning. To illustrate this, René highlights that with this technology, as much as 10’000 news streams can be retrieved per second for a network that generates 100 new content items (e.g., tweets) per second on a single machine – in comparison, Twitter reported that 600 tweets per second as of last year. This alone should be enough to emphasize the enormous potential of this new technology.
Archive for the ‘Development’ Category
Smartphone Apps – Curse and Blessing
Friday, July 30th, 2010Before owning an app-centric smartphone such as the Apple iPhone myself, my thinking has been that apps serve no real purpose – 90% of the purposes fulfilled by these apps exist as free Web applications and services, and app-centric smartphones usually come with capable Web browsers and data flat rate plans. My thinking has changed, however, since using an iPhone: apps have real (or at least, perceived) advantages when compared to Web applications accessed from a smartphone browser:
- User experience perfectly adapted to the phone (e.g. touchscreen interaction)
- Direct access to device “periphery”, e.g. gyroscopic sensor
- Integration with other software, e.g. messaging, contacts
- Own, colorful app icon
Auto-Starting EJBs
Saturday, June 13th, 2009An issue that recently gave me something to gnaw on was that of auto-starting an EJB: I wanted to achieve the effect that an EJB within a Java enterprise application would become active (and in that special case, connect to another component on another machine) as soon as it was deployed (or the container was started). If possible, I wanted the solution to be container independent. After a lot of googling and trying, I found a solution that works, but it is probably one of the most awkward hacks to solve a problem that sounds as simple as this.