TimesTen In-Memory Database
for Extreme Performance

Just how fast is TimesTen In-Memory Database?

Doug Hood
Evangelist for Oracle TimesTen, Oracle In-Memory and Oracle NoSQL

The TimesTen In-Memory Database provides excellent performance on even commodity hardware. RDBMS performance has two important factors, latency and throughput. Latency is about how fast SQL Select, Insert, Update, Delete or Merge operations can be completed. TimesTen is known for enabling really low latency SQL transactions. We measure TimesTen latency in microseconds rather than milliseconds:


This latency benchmark was run on commodity Linux / Intel hardware:

  • 2 CPU sockets, 22 cores/socket, Intel Xeon E5-2699 v4 @ 2.20GHz
  • The benchmark is TPTBM for Reads and Updates
  • TimesTen (100M rows of data, DB is 17 GB)

Having really low latency also helps with throughput. RDBMS throughput is defined in terms of (ACID) transactions per second.


This throughput benchmark was run on the same commodity hardware:

  • 2 CPU sockets, 22 cores/socket, Intel Xeon E5-2699 v4 @ 2.20GHz
  • The benchmark is TPTBM for mixed workload (80% reads, 10% updates, 5% inserts and 5% deletes)
  • TimesTen (100M rows of data, DB is 17 GB)

The performance of TimesTen is dependent on the workload, hardware and tuned SQL statements.  These TPTBM benchmarks used primary key lookups and the SQL statements in each transaction only affected a few rows.  If your SQL workload has complex joins involving many tables or returns a huge number of rows then these operations will take longer.  The hardware that the workload runs on also affects the performance.  Faster CPUs in terms of GHz, size of the L3 cache and number of cores tend to give better TimesTen In-Memory Database performance.

Download and try TimesTen In-Memory Database with your workload.  The TimesTen Quickstart Guide will lower the learning curve.

Disclaimer: these are my personal thoughts and do not represent Oracle's official viewpoint in any way, shape, or form.

Join the discussion

Comments ( 1 )
  • Vikram Sharma Tuesday, June 12, 2018
    Can we use TimesTen for decoupling an insert into a table which has high number of indexes ...by inserting a timesten layer having the same table (with or without indexes).

    My customer has some SLA in which the data can be made available for reads on oracle databases....
Please enter your name.Please provide a valid email address.Please enter a comment.CAPTCHA challenge response provided was incorrect. Please try again.