XDB: approXimate DB

XDB is a database that supports online aggregation using the wander join algorithm, based on the tranditional RDBMS database, PostgreSQL. It supports online aggregation query written in SQL with ONLINE keyword. It also features a query optimizer that optimizes wander join execution plans.

Source Code

Online Aggregation

XDB reports unbiased estimators and confidence intervals periodically for an aggregation query. The confidence interval shrinks over time. The true result Y is within the confidence interval with high confidence.

SQL Syntax

Online aggregation queries are written in SQL with a few additional keywords. The example query finds the total revenue loss due to returned goods in a TPC-H dataset. The online aggregation query will be run for 60000 milliseconds and results are reported every 1000 milliseconds. The confidence interval is computed with confidence of 95%.

PSQL Frontend

Online aggregation queries can be issued in a PSQL frontend and the results are reported at the end of the queries.

Apache Zeppelin Frontend

We also have a modified Apache Zeppelin frontend, which allows users to launch online aggregation queries in a web frontend and see the improvement of the results as the query is being executed.


          * authors in alphabetic order


  • Zhuoyue Zhao: zyzhao [at] cs [dot] utah [dot] edu (contact author)
  • Bin Wu: bwuac [at] cse [dot] ust [dot] hk
  • Feifei Li: lifeifei [at] cs [dot] utah [dot] edu
  • Ke Yi: yike [at] cse [dot] ust [dot] hk