The Oracle NoSQL Database Blog covers all things Oracle NoSQL Database. On-Prem, Cloud and more.

  • March 6, 2017

Oracle NoSQL Database performance on Oracle Cloud Infrastructure (OCI) IaaS

Ashutosh Naik
Senior Development Manager

We measured the performance of Oracle NoSQL Database release 4.3 running on Oracle Cloud Infrastructure (OCI) compute nodes using the YCSB workload. YCSB is the most popular benchmark for NoSQL systems. We are pleased to report the following performance results of our test.

Performance Benchmark Results



Insert Throughput

Mixed Throughput

Insert Latency (ms)


Read Latency (ms)


Update Latency (ms)



450 million








Hardware Configuration

We used three DenseIO nodes to run the Oracle NoSQL Database server and three Standard nodes for the YCSB driver application.

DenseIO node: 36 OCPU, 512 GB RAM, 28.8 TB NVMe SSD, 10 Gbps network.
Standard node: 36 OCPU, 256 GB RAM, 10 Gbps network.

Software Configuration

We used Oracle Linux Server release 6.8 on the client as well as the server nodes. We used Oracle NoSQL DB version 4.3, which can be downloaded from here. The server was configured as a 9 shard system with three-way replication (RF = 3).

Benchmark Configuration

We used the standard YCSB parameters: 13 byte keys and 1024 byte values. The database size was 450 million records. All insert and update operations were performed with a durability policy of simple majority. All read operations were performed with the NONE_REQUIRED read consistency policy.

A total of 6 client processes were used, with 2 client processes running on each of the 3 Standard Compute Capacity nodes. Each
client used 200 threads for the insert workload as well as the mixed read/update workload.

NOTE: Learn how Oracle assists developers by automating the installation of Oracle NoSQL Database on the Oracle Cloud Infrastructure (OCI). Click here for details.

Be the first to comment

Comments ( 0 )
Please enter your name.Please provide a valid email address.Please enter a comment.CAPTCHA challenge response provided was incorrect. Please try again.Captcha