To foster greater and more consistent use of the new 100 Gbps connections that is being deployed in the national RNP backbone, the e-Cyber project aims at delivering high-performing services to the most infrastructure-demanding research centers in Brazil. To do this, the project is getting inspired by the “superfacility” concept, which is adopted by initiatives like GRP (Global Research Platform) and EOSC (European Open Science Cloud). However, one of our biggest challenges is to engage the client institutions and bring them to co-create solutions and participate in the project governance.
5. e-Ciber project: conceptual superfacility model
RNP Academic
Network
e-Ciber network
High-capacity
High-performance
Scalable
400Gbps
100Gbps
Institution E
Institution D
Most infrastructure-demanding research centers in Brazil and supercomputing centers (“big consumers and big providers”)
Tailored to scientific network flows
Institution C
Institution B
Institution A
6. e-Ciber project: conceptual superfacility model
RNP Academic
Network
e-Ciber network
High-capacity
High-performance
Scalable
400Gbps
100Gbps
Institution E
Institution D
Most infrastructure-demanding research centers in Brazil and supercomputing centers (“big consumers and big providers”)
Tailored to scientific network flows
Institution C
Institution B
Institution A
+
new
member
institutions
signing up
7. e-Ciber project: conceptual superfacility model
RNP Academic
Network
e-Ciber network
High-capacity
High-performance
Scalable
400Gbps
100Gbps
Institution E
Institution D
Most infrastructure-demanding research centers in Brazil and supercomputing centers (“big consumers and big providers”)
+ Data Center
(High storage capacity)
Tailored to scientific network flows
+
new
member
institutions
signing up
Institution C
Institution B
Institution A
8. e-Ciber project: conceptual superfacility model
RNP Academic
Network
e-Ciber network
High-capacity
High-performance
Scalable
400Gbps
100Gbps
Institution E
Institution D
Most infrastructure-demanding research centers in Brazil and supercomputing centers (“big consumers and big providers”)
+ Data Center
(High storage capacity)
Pool of experts / Consultancy
for new projects and R&D
Secure cloud storage,
Co-location for HPC
Cibersecurity & Network
Operation Center
Service Layer
Tailored to scientific network flows
+
new
member
institutions
signing up
Institution C
Institution B
Institution A
Procurement negotiation with
suppliers / cloud services
R&D projects in collaboration
with academia and/or startups
“Trials” / data challenges with
vendors and startups
Workflow integration to
public cloud resources
9. Vision (moonshot)
Aiming for 10x improvement in the data-driven research
proccess
(in performance, usability, resilience and/or security)
Building a collaboration network of institutions keen to share knowledge
and computation resources in order to ensure readiness and predictability
of the computational resources needed for research.
Building a superfacility that will become essential for the most demanding
data-driven research in Brazil, able to influence public policies for
investments in ICT.
Additional goals:
10. Some initial hypothesis
• Customer facilities serve a community of researchers
that require reliable, secure and high-performance
remote access to research infrastructures.
• Customer facilities need access to additional reliable
computing power and/or data storage on demand,
with high performance and security.
• Central IT managers have a shortage of technical staff
to assist researchers to improve their data workflows.
• Customer facilities need specialized help to make
their research facilities more secure against cyber-
attacks.
• Customer facilities’ campus network are not ready to
support 100 Gbps throughput.
• Ensuring security is key for establishing trust between
institutions to enable resource sharing.
High CI-demanding
research centers
Researchers IT managers
C-Level
management
11. Layered model
Network Infrastructure
Campus network
Relationship management and sustainability
High capacity / perfomance Data Center
Monitoring + Security Operation Center
Service Discovery and Science Engagement
Deploy / upgrade Science DMZ architecture and DTNs.
Avoid bottlenecks inside the campus.
Identify and develop cross-facility workflows.
Identify and develop valuable services for Researches and IT managers.
Ensure customer's perception of value (benefits, usefulness, and importance).
Prospect funding opportunities / partnership with the private sector.
Active monitoring: network performance + cybersecurity.
High-capacity storage as-a-service.
High-performance co-location services.
Upgrade to 100Gbps all network path: backbone, metro networks and
last mile
Governance
Build a consortium of member institutions.
Develop a joint business operating model.
12. Layered model
Network Infrastructure
Campus network
Relationship management and sustainability
High capacity / perfomance Data Center
Monitoring + Security Operation Center
Service Discovery and Science Engagement
Ensure customer's perception of value (benefits, usefulness, and importance)
Prospect funding opportunities / partnership with the private sector
Governance
Build a consortium of member institutions.
Develop a joint business operating model.
I Know... We are in a Technology Exchange meeting.
You probably want to discuss something about the
usage of SDN, SENSE, Science DMZs, Jupyter, HDF5,
RUCIO, PerfSONAR, Federated Identity
technologies, Science gateways, Research Data
Repositories ...
13. …but there is a lot of lessons learned from other
initiatives
Superfacility@ LBLN, NERSC, ESnet and CRD Korea Research Platform
Pacific Research Platform
14. Main challenges / Issues
• How to succeed in the absence of substantial governmental investment?
(e.g.: no NSF or EC funding)
• How to engage institutions which are subordinated to different ministries and
belong to (very) different research areas? (e.g. Oil & Gas, Agriculture, Astronomy)
• How to integrate and work together with different Help Desk systems and teams?
• How to sustain a pool of experts?
• How to offer cybersecurity as a service to large research facilities?
• The role of a NREN in fostering a Superfacility.
15. e-Ciber project: Success metric
Infrastructure-only:
Last-mile capacity
upgraded to 100Gbps
Superfacility with
shared business model
(sharing and integrating computing resources,
knowledge, decisions, help desks...)
Initial Funding
CAPEX: Government funding (FNDCT) + RNP + contribution from
cooperation agreements on R&D (initially Petrobras and Embrapa)
• This is more than simply a model of connected research facilities at 100Gbps
bandwidth. The proposed model encompasses the full ecosystem of
infrastructure, services, software tools and expertise needed to make HPC
facilities secure and easy to use remotely.
16. Elements of a well developed Cyberinfrastructure
Ecosystem
Inspired by Colin Wright’s slide on Southern African Development Community (SADC) Cyber-infrastructure framework
Organizations
Research
Instruments
and facilities
Data
Networking
Software
tools and
middleware
Expertise,
R&D, Training,
Education
Interoperability
& operations
Computational
Resources,
Clouds, HPC,
Data Centers
Collaboration
Discovery
The main purpose of this presentation is o share with you our Superfacility Model and collect some feedback.
... And to make sure I’ll get some feedback I asked some friends with a lot of experience in supporting superfacilities to join this session and also share their thoughts
But not only them.
I would appreciate to get feedback from the audience. So, please feel free to jump in and comment.
TWO THINGS BEFORE WE START:
1. as eCiber just started, we still don’t have lessons learnerd. We have concerns. I will focus on the governance aspects
Main drivers for eCiber: the new 100G links and the need to choose the first institutions to be served by this new capacity