Download as pdf or txt
Download as pdf or txt
You are on page 1of 31

WELCOME

From Complexity to Simplicity


How We Built Single View at Chow Sang
Sang with MongoDB & Tapdata

TJ Tang
Founder & CEO, Tapdata
Agenda
Part I: Problem & Solution for Data Silos

Part II: Building Single View with Tapdata +


Mongo

Part III: Product Demo


About TJ

✓ Tapdata Founder

✓ Founding Engineer at A16Z founder Ben Horowitz’s Opsware, US

✓ Chief Architect at FedEx APAC, Singapore

✓ Principal Solutions Architect, Employee #1 at MongoDB Hong Kong


About Tapdata
60+ Enterprise Customers
$1.5M Angel funding $10M Pre-A Funding 10,000+ Cloud Users

2019.10 2021.09 2023.01

2019.09 2020.09 2022.07

Tapdata 2.0 GA
Founded in Tapdata 1.0 GA Open Source is released
Shenzhen,China
Part I:
The Modern Solution to the Data
Silo Problem
How Many Systems in Your Organization?

54% < 200 Systems

23% has 201~500 systems

15% has 501+ systems

F5 Application Systems Surey


QUESTION

How often do you looking in multiple places to get a


complete picture of a single customer?

How much time do you spend reconciling and


merging customer data from different sources?

How confident are you in meeting compliance


requirements when customer data is spread
across multiple systems?
WE TRIED!

Centralize into App Integration


Data Warehouse via Kafka

Stale Data! Complex &


Fragile
Real Time DaaS

An “Enterprise Cache” for your Mission Critical Data

Dashboards
EComm

PoS
Customer 360

CRM
Real Time Sync API Service Social CRM

CMS
Operational
Analytics
Help Desk DaaS Platform

3rd Party AI
WHY DAAS

Centralize, then Reuse Offload read workload to Fast data service,


modern, scalable data quick delivery
store
Part II:
Building Single View of Products
at Chow Sang Sang
About Chow Sang Sang

✓ Top Tier Luxury Company, nearly 100 years old

✓ More than 1000 shops in the region

✓ > HKD $20B Annual Revenue

✓ 27 Brands
Ever Growing Business Requirements
✓ Embrace Web, Social

✓ Third Party Social Platforms

✓ EComm Platforms

✓ 10s of thousands of Marketing


Campaigns each year
Data Challenges Faced by Team

◼ Siloed Data:
Product, inventory, orders, customers data scattered in dozens
of PoS/Warehouse/Product/CRM systems

◼ Difficult to Adapt/Scale RDBMS


20+ years in the making, complex
Performance unable to meet soaring demands

◼ Slow to Respond All Requests


When new apps require core data such as product, inventory,
often needs length process to collect and consolidate from
multiple systems
Single View of Product:
OmniChannel Product & Inventory

MICROSERVICES
Objective:
WEB

SYSTEM A
• OmniChannel Search
SHOPS

SYSTEM B
• Consolidated
DATA
Inventory
SYNC

SYSTEM C

BACK OFFICE
• HA & Scalable
MOBILE

SYSTEM D
Technical Considerations

◼ Find appropriate database solution for centralized data store

◼ Find ways to keep the data in sync between sources & centralized data store

◼ Ability to quickly create data models/views to serve changing requirements


Finding a Right Data Store

◼ Must be scalable, consolidating data


Sharding
from multiple sources
… Router Router

◼ Must be high performance & low In Memory
Performance Shard 1 Shard 2 Shard N
latency, serving API to web/mobile
Primary Primary Primary

◼ Must be versatile to store structured & Flexible


Secondary Secondary
… Secondary

Schema Secondary Secondary Secondary


semi-structured data
Finding the Right Data Sync & Processing Tool for MongoDB
Tapdata
Architecture

Full LOAD & TRANSFORM STORE & On demand


CONTINOUS SYNC On the fly CATALOG Service

OLTP DBS
OPERATIONAL APPLICATIONS

PLUGIN Framework Web

PRODUCT

SaaS/ Enterprise software Microservices


CRUD
ERP API Mobile
ORDER

CRM OPERATIONAL ANALYTICS

3rd Party APIs


INVENTORY

Structured data

Logs Reverse ETL

File and
Object CUSTOMER
STorage

Events
Step 1: Sync Changes in Oracle to MongoDB with CDC
Step 2: Merging from multiple sources to form consolidated Product

◆ Drag & Drop to merge data into one collection

◆ Optionally add transformation

◆ User Defined Function for complex processing

◆ Any changes in any source is auto-synced to the


target data model
Step 3: Accessing Unified Product Data via On Demand API

◆ Publish API Directly From Database

◆ Support MongoDB & Popular RDBMS

◆ Configuration Based, Instant Publish

◆ API Documentation
API Server / Atlas API
◆ API Permission

◆ API Stats
What started as a simple Oracle to MongoDB Replication
use case, 2 years later we achieved:
Results by Numbers

12 sources 5 days 5 apps


connected and data to deliver a data consuming product
sync-ed into API, compared to & customer data
centralized 4~8 weeks from DaaS
data platform
An Enterprise DaaS Platform

Product Catalog Discount Discount Analytics Reporting


New Applications
PG SelectDB

ES
/api/customer
/api/product

Push Sync RESTful API GraphQL Push Sync

Make Data Available


Master Data Model – Unified
CUSTOMER PRODUCT INVENTORY ORDER CHANNEL REF DATA

Consolidate Transform Compute Merge

Foundataional Data Model

Continuously Sync , Always Up-To-Date

Existing Applications
Shortened Cycle to Deliver Data API

Are data in N
DaaS ?

5
Y Sync Data From Sources

Y
Requires
Builindg View?

Build Materialized View


& Validate Data
N
2
Configure API
1

Test & Publish


2

Man Days
Part III: Product Demo
Turn complex into simple

- No Code Data Integration


- Flexible JSON Schema for Easy Conslidation
Sync from Oracle & Salesforce to Create Single View of Customer

1. Establish connections to Oracle DB & Salesforce

2. Sync customer table to FDM from Oracle table

3. Sync customer data to FDM from Salesforce

4. Merge these two tables into OmniCustomer

5. Add a JS processor to show case the UDF

6. Do update in Salesforce / Oracle to see changes in


Mongo
Summary

Why DaaS Why Tapdata + MongoDB

◼ Centralize for faster access ◼ Real time Integration

◼ Single View of Customers ◼ Flexible & scalable

◼ Single View of Products ◼ 4x Productivity Increase


THANK YOU

Try Tapdata for Free


In Cloud or On-Prem

Please come to meet us at our booth.

You might also like