EIQ Server® technical features of indexing and query processing

1. Extremely fast query response times. WhamTech’s query processing is executed “virtually” using indexes and Boolean operations on query result-sets. As an example of WhamTech’s complex query processing, Thunderbolt achieves sub-second responses to complex queries on a live one billion record database – see www.billionrecords.com for an online demo and details.

2. Very large databases (VLDBs). WhamTech was a VLDB technology company that dealt with VLDB issues; in particular, the combination of high performance of complex queries by a large number of users on VLDBs. EIQ Server brings many of WhamTech’s database technology benefits to other mainstream database technologies.

3. Real-time indexes with extremely fast update rates. WhamTech’s indexes allow INSERT/UPDATE/DELETE rates of up to 10s of 000s of records per second on a single server. Queries can be made on the indexes immediately following the brief moment they are updated. One proof-of-concept achieved a single-term query and insert rate of 80,000 records per second in a 60 GB, 300 million-record database on a dual-933 MHz Intel server with 4 GB RAM and local SCSI, 7200 RPM, RAID 5 disks.

4. Disproportionately high number of users per server. WhamTech’s RDBMS, Thunderbolt, and EIQ Server support a disproportionately large number of users per server due to:

   1. Extremely fast index and query processing

   2. Virtual query processing, where data in the database is not accessed until final result sets are isolated

   3. User channel reuse, where channels (or threads) to the database engine are available to other users before and after a query execution

5. Storage required for indexes only; data remains in the source database. WhamTech’s indexes are very efficient and usually require a lot less space than conventional indexes. Storage can actually be reduced overall if WhamTech’s indexes replace some of the source database indexes.

6. Unique WhamTech commands that allow multiple ways to present JOIN data, and a heuristic SELECT that performs link analysis and data mining functions that are traditionally dealt with through OLAP.

7. Immediate record counts…are available at every point in a query. These counts are automatically tracked at the data value level and are also available in result-sets. The data itself does not need to be counted.

8. Embedded Value Indexes™…are a separate form of indexes and can be used to accelerate access to low-level data, by storing higher or same-level data in low-level indexes. Value Indexes can be used to avoid or minimize extensive table-joins, compute aggregation statistics and aggregation data on the fly, e.g., SUM, AVERAGE, MAXIMUM, MINIMUM, MEAN, STD. DEVIATION, etc., and simplify complex queries. Value Indexes and normal indexes can be combined for ad hoc aggregations that would not be possible with other database index and query technologies.

9. Spatial and temporal queries…are the reason WhamTech was formed as a company to develop the database technology, which has exceptional range query processing. EIQ Server indexes are well suited to processing range queries, such as GIS and period data, without creating interim or temporary tables. WhamTech developed an oil and gas pre-processing utility that cuts down by several orders of magnitude the time it takes to organize raw seismic data, in particular, multi-component, 3D seismic data before processing and interpretation. This feature avoids the need to create OLAP-style PERIOD tables for subsequent analysis.

10. Listpick…is a means of only retrieving data that is needed at a specific time using the indexes or result-set pointers. For instance, if a user queries and isolates millions of records from a VLDB, the entire millions of records do not to be read and sent across a network to the user at one time, as this would take a long time, slow down the network and overwhelm the user. Instead, the user is only shown a limited number of records at any given time, e.g., 25, and is provided controls to navigate the list of data using pointers rather than actual data. This allows users to quickly page through huge numbers of records and jump to different points in the list of records without paging.