BDB Platform Performance Benchmark



BDB 4.0 Performance Snapshot

  1. Governed Dashboards Loading
    1. Web (4-5 Seconds)
    2. Mobile (5 – 7 Seconds)
  2. Self Service BI Loading
    1. Web (4-6 Seconds)
    2. Mobile (4-5 Seconds)
  3. AI-based Search Result (120 m Rows)
    1. Web (5-7 Seconds)
    2. Mobile (6-9 Seconds)
  4. Predictive Workbench Loading
    1. R Workspace (<1 Sec) | Python (<1Sec)
    2. Deep Learning (<2 Seconds)
  5. Data Wrangling
    1. Launching ETL <3 Seconds
    2. Launching Data Prep (3-5 Seconds)
    3. Profiling Data Prep (7-8 Seconds) (up to 10k rows)
  6. Data Pipeline
    1. Workspace Launching in 4-5 Seconds & Pipeline view in 3 Sec
Note: All Readings for a Business User using the standard Internet and standard Laptop for most common business workflow, for Server Configurations please see the Deployment Details >>

BDB Flow diagram

BDB Governed Dashboards Performance

Note: Dashboard loading times are highly dependent on ‘query optimizations,’ ‘dashboard designs’ and ‘network latencies. Please contact the BDB team for further optimizations.

BDB Self Service BI (Business Story)

Note: Business Story performance is highly dependent on the elastic store data volume upon which it is created.

BDB Platform

The following image displays the loading time of various modules and basic workflows in the BDB Platform 4.0.

AI-based Search (BDB Search) - (120 Million Records)

Below given is the detailed record of response time for Data Search Queries using AI-based Search for 120 Million records in an Elastic Data Store on BDB Platform 4.0

AI-based Search On different Mobile Networks

The details of the response time for AI-based Search on BDB Mobile Apps connected to the BDB Platform 4.0 are provided below. The readings are compared with various Network Bandwidths to check performance.

Predictive Workbench R workspace

The below given chart displays the Predictive Workbench performance in the R Workspace.

Predictive Workbench Python workspace

The below given chart displays the Predictive Workbench performance in the Python Workspace.

ETL & Data Prep Test

The below given image shows ETL-Basic Workflow with Large Amount of Data on 2 Core 16 GB Infrastructure

Data Preparation Transforms Performance Report

The following chart performance details of the Data Preparation Transforms.

Performance Report 4.0 - Tested Environment Details

The following table displays environment details in which the BDB Platform 4.0 was tested.

Application URL https://app.bdb.ai
OS Kubernetes Environment in Alpine OS hosted on Amazon Linux
BDB Meta Data DB MySQL 5.7 RDS
Java Version Java 8
Tomcat Version 7
Elastic Search Version 5.6.5
Karaf Version 3.0.7
Nginx version 1.2.2
Server Config (Platform & Elastic) Kubernetes Environment, Cloud Cluster
Allocated Resource 8 Core, 32 GB
Kubernetes Version : v1:10.3
Elastic Search 32 GB 4 Core * 3 Machines, AMQ 5.12
Browsers/Tools used for testing IE – 11, Chrome, Firefox latest Version, JMeter 3.2x