김기남_ATDC2016_160620_[키노트].key

Size: px

Start display at page:

Download "김기남_ATDC2016_160620_[키노트].key"

종옥 문
6 years ago
Views:

1 metatron Enterprise Big Data SKT Metatron/Big Data

2 Big Data Big Data... metatron Ready to Enterprise

3 Big Data Big Data

4 Big Data Big Data?? Data Raw. CRM SCM MES TCO Data & Store & Processing Computational Power Data Analysis Big Data Generate... value Insight and Action Sensor 40% ~ * Value leakage Value leakage Value leakage... * Source: McKinsey Digital

5 Big Data에 대한 오해 Better algorithms beat more data...?? 분석가들 사이에서도 갑론을박이 있지만 현재는 More data beats better algorithms!! ALGO RITHM DATA

6 Big Data Let the data do the work instead! Data driven shift Rule based Programming ( ) Machine Learning ( ) Data Data Algorithm Data Output Computer Computer Output Algorithm

7 Big Data Big Data...

8 Big Data BI R Visual Discovery Tool Big Data. + Visualization = Big Data Value? Finance Data CRM Data Operations Data More More Data Data R Visual discovery In-memory BI

9 Big Data DW 10T. BI Mart Raw Data Finance Data CRM Data Operations Data More More Data Data

10 Big Data Log Storage, ROI... Hadoop Computing Data Base system Computing Hadoop Computing CPU Storage 1 1 Storage CPU Storage 1 Storage CPU Storage Storage Storage X n ( ) high end ( ) low end ( )

11 Big Data Hadoop, RDBMS Tool Raw data Machine learning value Storage RDBMS (MPP) Analytics?

12 Introduction of metatron

13 metatron Core differentiations No Limits on Data Size Ready to Machine Learning No Coding

14 metatron Architecture Analytics Sub-second Processing Engine Storage & Processing Data Collection Big data Storage Automatic Data Collector & Pattern Detector A A A ONM(OPERATION AND MANAGEMENT) Advanced Analysis Functions (DNN, ML, etc) SECURITY Visualization RESTFUL API Visualization

metatron Focuses on Easy-to-use Data Preparation Process Automatic Data Collector Forwarder: OS Data type Real-time Ingestion Pattern Matcher: ( ) Forwarder Forwarder Forwarder

15 metatron Focuses on Easy-to-use Data Preparation Process Automatic Data Collector Forwarder: OS Data type Real-time Ingestion Pattern Matcher: ( ) Forwarder Forwarder Forwarder Aggregator Aggregator Aggregator Collector API Queue Visualiz ation Data Proce ssing Colletor API: visualization Queue: Drag & Drop Look-up Join 10,000 events/sec per core Pattern Matcher

16 metatron Processes time-series data in almost real-time without a size limit Sub-second processing engine Real-time Engine Advanced Analytics Query Engine Historical Engine Time-series Data index Search Aggregation (Sum, Average, Min, and Max) In-memory Historical data Processing Queue Real-time indexer Hadoop Distributed Storage Historical Engine Meta-store HDFS Storage Time-series Data Aggregation query : 1 rows/sec/8 core

metatron Provide Machine Learning and Statistical Analysis in distributed environment Advanced Analysis Functions Spark ML Lib Spark ML ML Distributed Machine Learning Supervised Learning

17 metatron Provide Machine Learning and Statistical Analysis in distributed environment Advanced Analysis Functions Spark ML Lib Spark ML ML Distributed Machine Learning Supervised Learning Classification Regression Feature Selection Unsupervised Learning Clustering Latent variable Models Signal Decomposition Distributed Statistical Analysis Time series and trend analysis Multivariate Analysis Anomaly detection Probability density Estimation Model Selection

18 metatron Fast visualize for massive data base on in-memory grids Visualization In-memory Grid Massive Data visual 100 Data 0.3,, Map(, Flash Map) Drill-down visual UI Component In-memory Grid (Buffered layer) Data Repository Report Dashboard Fast UI Component

19 Ready to Enterprise

20 Ready to Enterprise For Enterprise Architecture

21 Thank you

Intra_DW_Ch4.PDF

Intra_DW_Ch4.PDF The Intranet Data Warehouse Richard Tanler Ch4 : Online Analytic Processing: From Data To Information 2000. 4. 14 All rights reserved OLAP OLAP OLAP OLAP OLAP OLAP is a label, rather than a technology