This site offers the largest inventory of weekly vacation rentals in Cape May County, NJ!
This (example?) site was developed as a technique of a distributed http search method of different vacation rental web sites. Basically each site offers web-based vacation searching on their site and this 'master' site (and other 'regional' ones - OceanCityNJ.com, SeaIsleCity.com, ...) consolidates all sites into a uniform 'standard' search site.
Our experience in larger (3,000,000 records) 'static' database applications (ie. NJ.TaxRecords.com) and smaller (3,000+/- records) distributed rental reservation databases (ie. StoneHarbor.com, BergerRealty.com, OCNJ.com, ...) mandated a distributed search method for the more 'dynamic' vacation rental database applications of individual web sites. Each subscribing site handles a search request and responds in a 'fixed' way. This site requests, compiles, sorts, formats and displays the requests with hyper links to each respective 'subscribing' site for each property.
More exactly - There is no data stored on this site!
This distributed method would allow for 5-15 second searching of a vacation rental network of approximately 100,000 properties! Additionally, each subscribing member would maintain their own databases (and web site) and master and regional sites (like this one) would offer master 'one-stop' searching for all or part of the sites. Besides the benefits of speed, this would allow:
Perl and shell script CGI programs 'fetch' the data requested from each site and format it in 'ordered' html format. The program spawns a process for each requested site, waits a pre-determined amount of time (ie. 5 seconds), gathers each respondant site and compiles the results. A JAVA implementaion limits the processes to 1 but increases the need for 'threads'.
A reference implementaion of the Power Search engine has been developed using the Java programming language and its Servlet framework to help address some potential scalability issues that may appear while using the Perl/CGI method. Java was selected was because of its threading API and built-in network services. Utilizing threads of control in the PowerSearch engine serves a two-fold purpose. First using threads enabled concurrent searches of the desired domains while keeping one search site from stalling the entire search. Secondly, a threaded Java servlet approach allows for greater scalabilty of the application. In the Perl-CGI approach, the amount of server resources and child processes needed for searches grow linearly as the search sites and search clients increase. The threaded Java model will grow to no more than one process regardless of the amount of search clients or search domains. Although the size of the java process created to execute the servlet may grow large, because there is only one process to be monitored it should be easier to accomodate for that growth through the use of additional system RAM or faster processors.
Copyright © 1984, 2000 James S. Ginn
Another site by:
Attn: Lisa Haas
801 Asbury Avenue Suite 512
Ocean City, NJ 08226