This document summarizes a presentation about data virtualization and edge computing. It defines edge computing as performing data processing near the source of the data to optimize systems. Examples provided include analyzing factory data locally and treating individual countries as edges to comply with different privacy laws. The presentation discusses moving to an integration platform as a service (iPaaS) model that decouples the metadata layer from the execution layer, pushing query execution to local agents to reduce network traffic while still allowing centralized development and monitoring from the cloud.
Denodo DataFest 2017: Data Virtualization in the World of Edge Computing
1. O C TO B E R 1 9 - 2 0 , 2 0 1 7 N E W YO R K , N Y
Redefining Analytics for Successful Self-Service BI, Cloud, and Big Data
The Agile Data Management
and Analytics Conference
#DenodoDataFest
2. Data Virtualization in the world of Edge Computing
Pablo Alvarez
Principal Solution Architect, Denodo
4. Edge computing is a method of optimizing computing
systems by performing data processing at the edge of the
network, near the source of the data.
What is Edge Computing?
6. 6
Why Edge Computing?
▪ Lower Cost
▪ Reduced storage, traffic and processing in the central hubs
▪ Minimize latency
▪ Data is analyzed at the local level
▪ Reduced network traffic
▪ Only necessary data is transmitted to the central hub
▪ Two tier access for better performance
▪ Critical applications can access the Edge directly instead of the central hub
8. 8
How far does it makes sense to push the edge?
▪ What is your business problem and
where can you benefit from the
ideas of Edge Computing?
▪ Keep in mind that Data
virtualization is a data integration
and data delivery technology
▪ What’s the global architecture?
▪ Cloud vs data center
▪ Geographical distribution
▪ Etc.
Source:
https://medium.com/iotforall/whats-proximity-computing-4b713ecf3666
9. 9
Example: Country as the Edge
▪ Multinational company with
independent regional units
▪ Have to comply with multiple
data privacy laws
▪ Need for centralize world-wide
perspective
Your Edge is the country/region
Two-tier Logical Data Warehouse
Architecture
USA data center Europe data center
10. 10
Example: Factory as the Edge
▪ Multiple factories with locally
replicated systems for some
operations (supply chain, HR,
IoT manufacturing systems,
etc.)
▪ Large amount of data
generated at the factory level
▪ Huge potential for distributed
processing for aggregated
global reporting
Your Edge is the Factory
12. An Integration Platform as a Service (iPaaS) is an data
integration platform that combines the benefits of a PaaS
platform (cloud hosted) with the performance of local
processing via execution agents
What is iPaaS?
13. 13
Beyond Denodo 7: Denodo’s iPaaS Architecture
Execution Agent Execution Agent
Metadata
Repository
Execution Engine
& Optimizer
Data center A Data center B
Corporate Security
Monitoring &
Auditing
“The Cloud”
14. 14
iPaaS model and Edge Computing
Decouples metadata layer (Cloud) from the execution layer (on-prem):
▪ Data sources, view definitions and usage info are stored centrally in the cloud
Actual query execution can be pushed down to local “execution agents”
deployed at the edge to avoid heavy network traffic
Simplifies operations:
▪ Development is done directly in the cloud via a web tool
▪ Components are either web-based (clients) or automatically updated (agents)
▪ Execution agents are monitored centrally from the cloud UI