raymondtlin / ingestd-tpcdi

Data Integration via Confluent Kafka

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

ingestd

The TPC-DI Benchmark

Created by the Transaction Processing Council as a benchmark for ETL applications, the TPC-DI benchmark involves integrating multiple data sources and file formats into a data warehouse schema. The benchmark specifies requirements both for historical and incremental loading, as well as transformations to be applied.

Dataset

As outlined in the official TPC documentation found here, the data model for the benchmark is a retail brokerage firm.

Flat files originate from 5 primary sources:

  1. OLTP: CDC extracts, which are bar-delimited txt
  2. HR: comma delimited database extract
  3. Prospect List: comma delimited csv
  4. Financial Newswire: Variable Fixed Width Format
  5. CRM: XML

The destination DWH schema will have the following:

Facts Dimensions Reference
fCashBalances dCustomers TradeTypes
fHoldings dAccounts StatusTypes
fWatches dBrokers TaxRates
fMarketHistory dSecurities Industries
fProspects dCompanies Financials
fTrade dDate, dTime

Kafka Ingestion & Integration

Data Serialization, Versioning

About

Data Integration via Confluent Kafka

License:GNU General Public License v3.0


Languages

Language:Python 100.0%