Semantic Technologies for Enterprise Cloud Management
1. Peter Haase, Tobias Mathäß, Michael Schmidt,
Andreas Eberhart, Ulrich Walther
fluid Operations AG
Semantic Technologies
for Enterprise Cloud Management
ISWC, November 11, 2010, Shanghai
2. Motivation
• Cloud Computing as a model in support of
„everything-as-a-service“
• Several benefits for the consumer
• Sold on demand
• Elastic
• Fully managed by provider
• Private clouds becoming increasingly important
• Enterprise-internal virtualization
• Can be linked to public cloud solutions
• Scalable access to computing resources and IT services
vision: fully automated data center
3. Enterprise Clouds – the eCloud Vision
All resources of an adaptive, cloud-enabled IT environment can be set up,
monitored, and maintained from a single, unified, and intuitive management
console:
Internal and external IT resources accessible across stack without vendor lock-in
High degree of automation and IT provisioning at click of button on the level of enterprise
landscapes
Internal portal of private/public IT services with e.g. pay-as-you-go cost models
4. Manage IT like an eCloud
Stack virtualization
and semantic
integration as
foundational
capabilities for
efficient automation
CXOsIT admins Application customers
Different user groups
with diverse demands:
administration,
documentation,
reporting, analysis, …
5. Challenge 1:
Data Integration
MonitoringandManagement
ApplicationTemplates
Hardware Layer
Landscape Layer
Virtualization Layer
Network Computing ResourcesNetw.-Att. Storage
V
L
VLM
VL VLM
VL VLM
VL VLM
• Awareness of full IT
stack required,
from storage to
application layer
• Heterogeneity of
resources across
layers of IT stack
• Heterogeneity
across different
vendors and
product versions
6. Challenge 1:
Data Integration
MonitoringandManagement
ApplicationTemplates
Hardware Layer
Landscape Layer
Virtualization Layer
Network Computing ResourcesNetw.-Att. Storage
V
L
VLM
VL VLM
VL VLM
VL VLM
• Awareness of full IT
stack required,
from storage to
application layer
• Heterogeneity of
resources across
layers of IT stack
• Heterogeneity
across different
vendors and
product versions
Use semantic data model for integrating semantically heterogeneous
information to get a complete picture of the entire data center
7. Challenge 2:
Collaborative Documentation and Annotation
• Technical base information retrieved
automatically from provider APIs
• Challenges
• Free-text documentation and augmentation of technical data
• Associate bussiness information with technical data
• Address heterogeneous data in a unified way
• Use Cases
• Which gold-level customers are affected if a storage filer breaks?
• Which resources did department X consume within the last months?
8. Challenge 2:
Collaborative Documentation and Annotation
• Technical base information retrieved
automatically from provider APIs
• Challenges
• Free-text documentation and augmentation of technical data
• Associate bussiness information with technical data
• Address heterogeneous data in a unified way
• Use Cases
• Which gold-level customers are affected if a storage filer breaks?
• Which resources did department X consume within the last months?
Apply Semantic Wiki technology to support collaboration
9. Challenge 3:
Intelligent Information Access and Analytics
• Different user roles with varying information needs
• Administrators
• Which resources am I responsible for?
• What underlying components may cause application X to freeze?
• Which IP addresses are currently in use?
• Customers (service consumers)
• What is the status of my systems?
• Which projects am I involved in?
• CXOs
• Which compute resources are currently available?
• What is the average CPU load of all VMs running on host X?
10. Challenge 3:
Intelligent Information Access and Analytics
• Different user roles with varying information needs
• Administrators
• Which resources am I responsible for?
• What underlying components may cause application X to freeze?
• Which IP addresses are currently in use?
• Customers (service consumers)
• What is the status of my systems?
• Which projects am I involved in?
• CXOs
• Which compute resources are currently available?
• What is the average CPU load of all VMs running on host X?
Expressive ad-hoc queries that overcome the border of data sets.
Visualization and visual exploration tools for structured data.
11. Our Solution:
Widget-based UI
• Resource-centric presentation
• Living UI, which exploits semantics
of underlying data
• Large collection of predefined
widgets, easily extendable
Search and information Access
• Coexistence of structured and
unstructured data
• Different search paradigms
Data integration through providers
• Convert data from a data source
into RDF data format
• High degree of reusability
• Customizable, easily extensible
12. Unifying OWL Data Model
Extract of the eCloudManager Intelligence Edition data model
13. Data Integration by Example
Predicate
Subject Object
Predicate
Object
Predicate
Predicate
Object
Predicate
Object
Object
Object
Subject
Predicate
Predicate
Object
Subject
Predicate
Object
EMC Storage
Provider
Data Provider Layer
14. Data Integration by Example
Predicate
Subject Object
Predicate
Object
Predicate
Predicate
Object
Predicate
Object
Object
Object
Subject
Predicate
Predicate
Object
Subject
Predicate
Object
EMC Storage
Provider
Data Provider Layer
Subject
Predicate
Object
Predicate
Predicate
Object
Predicate
Object
Object
Object
Subject
Predicate
Object
Virtualization Software
Automatical alignment by
flexible, key-based
generation of unique URIs
for the same components
across different providers
vmware
Provider
15. Collaborative Documentation and Annotation
• Technical Documentation
• Resource-centric view
• Edit wiki pages associated with data center resources
• Interlinkage of Resources
• User-defined Semantic Links in the Semantic Wiki
• Completion of missing data
• Ontology-driven edit forms
Wiki Page in Edit Mode … … and Displayed Result Page
16. Flexible, Living UI
• UI flexibly adjusts to semantics of underlying data
• Which widgets to display for a resource depends on its properties
• UI thus automatically composed based on the semantics of the
underlying data
• Widgets with varying functional focus
• Visualization (e.g., PivotViewer)
• Navigation (e.g., browsable graph view)
• Collaboration (e.g., Semantic Wiki pages)
• Mashups (e.g., connected product catalogs)
17. Search and Querying
• Coexistence of structured and unstructured content requires
hybrid search
• Different search paradigms
• Simple keyword search
• Structured queries using SPARQL
• Form-based search
• Faceted Search
• Query translation
diversity covers different use cases and user groups
18. Dashboards, Analytics, Reporting
• Queries can be directly included into Wiki pages/templates
-> considerably lowers effort in maintaining Wiki
• Evaluated dynamically when user visits the Wiki page
• Type-based template mechanism
• Visualization of queries as
• Table Results
• Bar Diagrams
• Time plots over
historical data
• …
Stacked Chart: Virtual Machines over time grouped by status
19. Ad-hoc Data Exploration
• Leverage Pivot Viewer for Linked Data
• Set-based exploration of heterogeneous resources
• Integrated view on techical and business-level resources
• Filtering with
faceted search
• Grouping by
different aspects
Visual data exploration with the PivotViewer
20. Experiences and Lessons Learned
• RDF-based data integration approach with provider concept
brings significant advantages in heterogeneous environments
• Flexible, easily extendable
• Fast setup (typically less than one day for new data centers)
• Integration of additional data sources unproblematical
• Semantic Wiki brings many benefits
• Step from Wiki to Semantic Wiki feasible
• Integration of live data (tables, charts, timeplots, etc.) in Wiki
perceived as great benefit
• Fast customization often replaces development of new modules
21. Experiences and Lessons Learned
• Positive feedback on novel interaction paradigms
• Visual exploration with Pivot viewer offeres unprecedented user
experience
• Graph view to better understand connections between resources
• Semantic Technologies scale well to large data centers
• For large data centers few millions of RDF triples
• Aggregation of historic data to keep dataset manageable
• Particular technical challenges we had to address
• Scalability: take care on how you do it!
• Missing features in current SPARQL implementation
• Aggregation
• Annotations
22. Related Projects
• Benefit: high reusability of underlying technologies
• Generic technologies for data integration, search, exploration etc.
• Can seamlessly be applied to other domains
• Core technologies of eCloudManager Intelligence Edition
available as Open Source Platform for self-service Linked Data
application development:
Visit our
• Linked Open Data demonstrator and
• Life Science demonstrator
at http://iwb.fluidops.com!
The Information Workbench is publicly available as Open Source project
23. Thank you for your attention!
CONTACT:
fluid Operations AG Email: info@fluidOps.com
Altrottstr. 31 Website: www.fluidOps.com
Walldorf, Germany Tel.: +49 6227 3849-567
Interested in more information?
Then check out our Information Workbench brochure in your ISWC 2010 starter pack!