JustOneDB is Fast
Rows can be rapidly inserted into tables – much like a row store without any indexes at all. Yet simultaneous queries provide fast and immediate access to that same data – much like a row store with every possible column indexed. All of this happens in a familiar transactional SQL environment.
Benchmarks on like-for-like hardware demonstrate that JustOneDB performs well over 100 times faster than a conventional RDBMS with the same functionality
There’s no need to perform unnatural acts to achieve great performance – you don’t need to partition, shard or compromise on consistency to achieve stellar performance. Moreover, you don’t even need to design any indexes to achieve great query performance.
JustOneDB Handles Big Data
Tables with hundreds of billions of rows are effortlessly handled on a single server – providing fast transactional access to each and every row. Modest hardware platforms using inexpensive storage systems can deal with billions of transactions per day and can retain that information indefinitely.
There is no need to scale-out unnecessarily just to overcome the inadequate performance of tired database technology.
Example Customer Application
The following provides an overview of performance results achieved within a high velocity, high volume customer application.
Hardware Platform Used
The results were achieved on a single server using the following modest hardware configuration:
Observed Insert Rate
The load rate below represents the observed insertion rate for 10 billion records of data inserted into a single table which has 22 columns containing number, text and date data with a range of different values and cardinalities. This single table represents 1.6TB of data but is neither partitioned nor distributed. All rows were inserted using one session (single thread) with transactions committing at every 50,000 rows.
The following table summarizes the average insertion performance across all 10 billion rows.
Concurrent Query Performance
The following table shows the average query time for a range of selective queries across 10 billion rows (1.6TB) with concurrent insert activity running at full velocity.
A unique aspect of JustOneDB is that no indexes are used yet queries perform as though every column were wholly indexed. The following table illustrates the ability of the database to support arbitrary SQL queries against the table without any indexes being used. Again, these queries were run with concurrent insert activity running at full velocity. Each query is shown as pseudo SQL code for clarity.