SDIC'16 - Betrieb des Smart Data Innovation Labs - Vorstellung der Plattform
-
Upload
smart-data-innovation-lab -
Category
Data & Analytics
-
view
114 -
download
0
Transcript of SDIC'16 - Betrieb des Smart Data Innovation Labs - Vorstellung der Plattform
KIT – Universität des Landes Baden-Württem ber g undnationales Forschungszentrum in der Helmholtz-Gemeinschaft
Steinbuch Centre for Computing
www.kit.edu
Betrieb des Smart Data Innovation LabsVorstellung der Plattform
Nico Schlitter, 13. Oktober 2016
2 13.10.2016 Betrieb des Smart Data Innovation Labs (SDIL) Steinbuch Centre for ComputingSteinbuch Centre for Computing
Partners
3 13.10.2016 Betrieb des Smart Data Innovation Labs (SDIL) Steinbuch Centre for ComputingSteinbuch Centre for Computing
SDI-X funded by BMBF 09/2015 – 08/2018to strengthen the cooperation between industry and science
self organizing data innovation communities industry research partner meet at community meetings and define projectsSDIL strategy board approves project proposals within two weeksproject members get access to the SDIL platform hosted by KIT
Smart Data Innovation Lab
MedicineSmart Cities
EnergyIndustry 4.0
4 13.10.2016 Betrieb des Smart Data Innovation Labs (SDIL) Steinbuch Centre for ComputingSteinbuch Centre for Computing
Running the SDIL platform at KITDevelopment of tools, best practices und operating conceptsDeveloping a legal frameworkResearch in the field of
data processing infrastructuredata integrationdata analyticsdata curationdata anonymization and pseudonymization
Tasks within SDIL
5 13.10.2016 Betrieb des Smart Data Innovation Labs (SDIL) Steinbuch Centre for ComputingSteinbuch Centre for Computing
funded by the state of Baden-Württemberg 10/2014 – 09/2017educating SME in regard to data analytics (~30 contacts to far)free of charge
consulting for local SME in the field of data analyticsevaluating the potential of existing dataperforming data analysisrecommending additional data collectioninvestigating further analysis steps
Smart Data Solution Center BW
• First contact• Checking requirements• Defining Objectives• Legal Issues
• Data hand over• Data analysis• Evaluation
• Presenting results• Recommendations• Next steps
Preparation Realization Finalization
6 13.10.2016 Betrieb des Smart Data Innovation Labs (SDIL) Steinbuch Centre for ComputingSteinbuch Centre for Computing
SDIL Platform Overview
7 13.10.2016 Betrieb des Smart Data Innovation Labs (SDIL) Steinbuch Centre for ComputingSteinbuch Centre for Computing
#Cores RAM [GB]
Disk Space [TB]
Network Software
IBM Watson FoundationPower 8
7 x 20= 140
4096 300 40Gbit/sEthernet
IBM Open Platform with Hadoop/SparkSPSS Modeler
SPSS Analytic ServerDB2 with BLU Acceleration
SAP HANA 4 x 80= 320
4096 80 10Gbit/s Ethernet
SAP HANAPredictive Analysis LibraryBusiness Function Library
Software AGTerracotta
( * on request * ) BigMemory Max
HTCondor 32 x 4= 128
1024 1Gbit/s Ethernet
RapidMiner, Python, R, Matlab,
Virtualization 3 x 12= 36
576 6 10Gbit/sEthernet
Red Hat Enterprise Virtualizationon GlusterFS
SDIL Platform Resources
8 13.10.2016 Betrieb des Smart Data Innovation Labs (SDIL) Steinbuch Centre for ComputingSteinbuch Centre for Computing
security is important due to storing of sensitive datastrict access control of the server roommultiple state-full firewalls are in placeaccess via dedicated login serverfine granular permission management for all SDIL components
unified storage solution enables users to use several different technologies w/o the need to migrate large volumes of datahome and project directories with snapshotsencrypted backup on tape for disaster recoveryticket system (xGUS) for user support
SDIL Operations – user perspective
9 13.10.2016 Betrieb des Smart Data Innovation Labs (SDIL) Steinbuch Centre for ComputingSteinbuch Centre for Computing
legal issues are much bigger than expected
heterogeneous infrastructuredifferent architectures and operating systemsspecial software stacks like HANA and Terracottanot everything is nicely packaged
data privacy protection
SDIL Platform Challenges
10 13.10.2016 Betrieb des Smart Data Innovation Labs (SDIL) Steinbuch Centre for ComputingSteinbuch Centre for Computing
3 networks: BMC, deployment/configuration, servicesTwo firewalls, strict access control, encryption IBM Spectrum Scale as storage solutionForeman and Puppet for deployment and configuration of SDIL machines (continuous integration)GitLab as code repository
Icinga for monitoring (integrated with puppet)Redhat Enterprise Virtualization (RHEV) on glusterFS
Ticket system for all operation tasksWeekly shifts for incident response
SDIL Operations – nerd perspective
11 13.10.2016 Betrieb des Smart Data Innovation Labs (SDIL) Steinbuch Centre for ComputingSteinbuch Centre for Computing
IBM Watson Foundations on Power 8
12 13.10.2016 Betrieb des Smart Data Innovation Labs (SDIL) Steinbuch Centre for ComputingSteinbuch Centre for Computing
SAP HANA & RHEV
13 13.10.2016 Betrieb des Smart Data Innovation Labs (SDIL) Steinbuch Centre for ComputingSteinbuch Centre for Computing
Continuously upgrading …