# Kafka use cases - April 29th - ThaiBev-CMKL Enterprise Architecture

## Kafka-services:

* Sales
* Domestic
* Partner
* Logistics

## Details
### Pae (Sales):

* Pubsub / Replicate
* 1-2 M per message
* < 500 M message / s (peak) / Burst
* Site 2 site

### FTD (Cash to delivery) - (transactional support)

* Database sync (burst)
* Data exchange between services (pub/sub) near real-time (3-5 sec)
* Order request (buffer) - message count
* User fire message every 10 second
* 500 users

[Client - exactly once]

### Vee FTD

* DB/SIEM - data replication ? ETL (Persistence)
* [Kafka stream] -> sizing
* Db-creation 200,000 record per transaction
* Monitor queue status (queue size)

### DR/no-DDD (Yu)

* Agile RDR

### Procurement

* Docker - end/document
* Agile ppp async/material? Sync with SAP
* Data sync with SAP (100k - 1M record) - (1 topics per project)
* Event storage (backup) / transactions dump from SAP
* For data source repository
* (Eventually change from cross-fire / agile) - split to transaction / sync topics

## Jiab (Approval)

* Replicate / DB (Daily sync + incremental sync 5-10 min)
* Service buffer (approval app) - service required to build workflow (queued) (a few seconds) - 100 tx/m (with peak on working hour, potential burst afternoon)
* Pub/Sub - Data publishing for other teams

***Front-zone -> transform to reporting zone / reporting database

* 2am-7am (only once a day to transform record)
* Sales 1M record per day (with ETL) - join on the fly
* (Want near real-time x Join)

## Dong

* Trigger*
* SQL R/W - messaging for event sourcing to split read & write database
* Use for replicate master data (both daily job + user event on demand monthly)
* 200 message / day <—> trigger

## Dew

* Author / event / seminar / concert / mini-concert (millions request burst)
* 500 tickets - 20,000 tickets per event (in 1 minute)
* Small message type (for transaction)
* Common application (mailing - waiting for the log)

## Yu

* Employee data publishing for other teams
* Replicate (100 k record)
* Low volume 1000 tx per day
* Check-in / Check-out 8.30-5.30 (burst - employee)
* Sync with success factor

## FTD

* Tx per day = DBceim sync to reporting = 10000 record per day / update 5 record per day = 50000 tx per day
* Transaction order: Bursted for 1pm-5pm

Use cases

* Trigger
* Use for sync to reporting
* Publish data
* Transaction support (near real-time)