Presentation is loading. Please wait.

Presentation is loading. Please wait.

DBOS DecisionBrain Optimization Server

Similar presentations


Presentation on theme: "DBOS DecisionBrain Optimization Server"— Presentation transcript:

1 DBOS DecisionBrain Optimization Server
Filippo Focacci Michel Eisenmann

2 Leading Developer of Advanced Analytics Solutions
We implement advanced analytics solutions to help companies reach their business operation goals THOUGHT LEADERSHIP IBM PARTNERSHIP GLOBAL PRESENCE 30+ years experience in advanced analytics Robust growth trajectory with over 30 employees, most PhDs Largest global IBM business partner for optimization solutions Beacon Award finalist in 2018 France Hong Kong

3 MANUFACTURING / LOGISTICS
Global Partners Who Trust our Advanced Analytics CUSTOMERS WORKFORCE DEPLOYMENTS MANUFACTURING / LOGISTICS SUPPORT TO R&D

4 A Platform Built to Ease Optimization Development & Deployment
A lean but powerful system designed to help you build and deploy fully scalable applications – quickly, easily, efficiently. MODULAR ARCHITECTURE Comes with pre-embedded workers and support for optimization models using CPLEX or OPL STATE-OF-THE-ART TECHNOLOGY Relies on standards and open source software, supports both local (Docker) and enterprise (OpenShift) configuration EMBEDDED MONITORING CAPABILITIES Simple web console makes it easy for users and administrators to monitor real-time job execution, KPI evolutions, and manage queuing and failovers

5 Main Concepts Easy–to-access from a Web Console or an external Client (Rest API) MAIN ARCHITECTURE Client Web Console Master Optimization Worker Tasks: - CPLEX - CPO Task: OPL Database MongoDB Messaging Rabbit HQ DB Optimization Server RUN & MONITOR JOBS A job has a type (Task) and input/output One worker may support multiple Tasks

6 DBOS Key Features Interactive Web Console
Off-the-shelf support for IBM ILOG CPLEX Docker-based deployment Extensibility and full customization Support for Kubernetes/OpenShift Integration via REST API Replay Executions OR Department Development teams IT Department Dev/Deployment Teams

7 Web Console – Drag-and-drop interface
Create and start a new job (from task definition) QUICK JOB CREATION

8 Web Console – Job Monitoring
Track KPIs, progresses and logs in real-time MONITOR JOB EXECUTION

9 Web Console – System Monitoring
REAL-TIME UPDATES Quickly review system behavior Immediately track execution issues

10 Web Console – System Monitoring
REAL-TIME UPDATES Quickly review system behavior Immediately track execution issues

11 On-shelf CPLEX/OPL workers
Pre-defined workers for CPLEX/OPL* to minimize processing time and maximize efficiency CPLEX / CP OPTIMIZER WORKER (Integer) Linear Programming Quadratic Programming Constraint Programming OPL WORKER CPLEX/OPL WORKERS Web Console Master Optimization Worker Tasks: - CPLEX - CPO Task: OPL Database MongoDB Messaging Rabbit HQ DBOS For OR: Thanks to DBOs, they can use any optimization language, they want to run the optimization -Easy to run optimization INPUTS PROCESS OUTPUT Optimization model Data & settings Solution *version 12.8+

12 Docker-based Deployment
As efficient as virtual machines, but much faster and modern with small footprints on servers, making it deployable everywhere. DOCKER PACKAGING Client Web Console Master Optimization Worker Tasks: - CPLEX - CPO Task: OPL Database MongoDB Messaging Rabbit HQ DBOS Local with Docker-compose Quick and simple Template available DEPLOYMENT Speaks to IT guy

13 Easy to extend and customize
Have the flexibility to design custom workers using Java-based code CUSTOM WORKERS Custom Worker Your Task(s) DEFINE AND IMPLEMENT TASK(S) WITH: Optimization Server Web Console Your optimization code Your machine learning code Any kind of CPU intensive computation Master Messaging Rabbit HQ Optimization Worker Tasks: - CPLEX - CPO Speaks to Developer Database Optimization Worker Task: OPL MongoDB

14 Example: Simple Java Client
Speaks to Developer

15 Example: Simple Custom Worker
Speaks to Developer

16 Web Console - Ready to Support Extensions
CLICK-TO-RUN No web development required to support extensions IMMEDIATE DATA PREP Dynamically transforms and prepares data no matter its format or complexity

17 Designed for Advanced Deployment
Supports both an industrial deployment (full Kubernetes and OpenShift compatibility) and local deployment using Docker-Compose DEPLOYMENT Cluster with OpenShift Web Console Master Optimization Worker Database MongoDB Messaging Rabbit HQ Server Failover Easy scalability Easy automation Resource monitoring IT guy Quick and simple Template available Local with Docker-compose

18 Seamless, Easy Integration
Use of REST API technology makes it easy to integrate with projects involving complex mathematical modeling MULTIPLE LANGUAGES / FRAMEWORKS SUPPORT Developer persona : enables them to use multiple languages (focus on this) Not importatn for IT guy

19 Replay Job Executions – Monitor and Retrieve
Client Web Console Master Optimization Worker Database MongoDB Messaging Rabbit HQ Server Replay DOWNLOAD LOGS & REPRODUCE Monitor Failures Retrieve input conditions

20 Map-Reduce for Optimization: Build and Deploy Optimization Decomposition on Distributed Resources in the Cloud  Today 11:00am-12:45pm Room 306


Download ppt "DBOS DecisionBrain Optimization Server"

Similar presentations


Ads by Google