Have you tried Dieselpoint Search™ yet?

 

Scalability

Current Dieselpoint customers are searching hundreds of millions of documents and hundreds of gigabytes of data. Performance requirements in some cases exceed 100 searches/second per server. In all cases the typical search is returning results in under 500 milliseconds.

The system supports near-real-time searchability of data – in other words, data becomes searchable only a few seconds after being added. One customer is currently using this feature to advertise hot, limited-availability music CDs at special prices. Once the CDs go on sale, there is a rush, and then the special must be removed once all items have sold out. This isn’t possible with a typical search engine.

Dieselpoint Search can scale linearly by adding additional inexpensive servers.

Indexing speed (on a single server) runs from thousands of items per second for database records to a tens of items per second for documents like MS Word or PDF. The size of the index is typically smaller than the size of the raw data, and often it is much smaller. This gives Dieselpoint a speed advantage over search engines whose index sizes are sometimes 2X or 4X the size of the raw data.

Search performance is typically measured using three numbers: the number of queries handled per second, the average search time in milliseconds, and the number of gigabytes of data handled. Depending on the complexity of the query, Dieselpoint can handle from tens to hundreds of queries per second on a single server over multi-gigabyte indexes. Typical query response time is from ten to a few hundred milliseconds. By adding additional inexpensive servers in parallel, Dieselpoint Search can achieve thousands of queries per second.