If we look at implications how our Information Ranking affects database size and performance, a good standard reference can be any commercial off-the-shelf (COTS) hardware, using a relatively simple contemporary PC Intel-based hardware server with about 4GB RAM, 1Ghz CPU and plenty of disk space, for example, some 80GB hard disk. It can serve either as a stand-alone database server or as a cluster node for Clusterpoint DBMS platform software.
Here are ball-park figures. You can effectively manage and search up to 5-10 million office and Web-size documents or up to some 100-200 millions of smaller XML / JSON documents (e.g., your SQL database records, exported in XML / JSON and stored into Clusterpoint database) per single server or cluster node, run on a very simple our sample reference server hardware installed with Clusterpoint DBMS software, and delivering sub-second search response times with disks access involved. Naturally, increasing hardware capacity, gives also better performance and enables to increase the number of stored documents in the database storage. As Clusterpoint Server software is designed to take advantage of all processing power, RAM and storage capacity, this translates also into very good vertical scalability (scale-up vs scale-out) of Clusterpoint DBMS architecture.
For larger amounts of information you should better use cluster database configuration, installing multiple Clusterpoint Servers and configuring a single "striped database" which is a distributed database on a system level (Please see Clustering). If you need to run at really high speed, you can partition your database even among larger number of hardware nodes, increasing individual node performance to the required level of maximum response times. In this way you can scale search response time to as low as necessary (for in-memory databases cached by OS and tuning, easily achieving <0,005 seconds), also achieving 1/Nth decrease for the total database volume indexing time. Do not forget, that for large databases short indexing and re-indexing time could be as important criteria as search performance, especially, if those databases are servicing 24/7 mission-critical applications. Clusterpoint XML- or JSON- only database platform provides both types of scalability in a cluster: for search performance and for flexible database indexing, empowering our customers with a flexible data storage and management platform that scales workload according to the business requirements.
Clusterpoint Information Ranking is core feature that delivers this flexible data base scale out ability in Clusterpoint architecture. You should always apply Information Ranking rules for your massively scalable databases, and the rest will be done by Clusterpoint Server. It will automatically span out all index creation and pre-sorting tasks and perform search consolidation tasks during search queries, among all available hardware in a cluster. This is a nice feature, which comes with Clusterpoint Server software and does not require to partition database at application level.
With overall Information Ranking applied as described and illustrated above, any of your database becomes scalable in a massively clustered database storage environment, without negatively affecting database search performance, and still continuing to provide relevant and fast search from customer point of view. The resulting customer web application system that uses Clusterpoint Server software as its core DBMS data storage, access and search infrastructure software facility, delivers guaranteed sub-second search performance in a managed transaction processing environment, required to run large scale mission-critical database applications.
Please read also sections Clusterpoint Indexing and Clusterpoint Search.
