International Options How can we scale Jennifer M. Schopf, PhD Director, International Networks August 8, 2017

International Networks @IU Links

But its about more than just the network Measurement and Monitoring Is the link working? Who’s using it? Are they doing ok? Working with end users to support science! Better data access == Better science Who do you ask when you data transfer isn’t right?

But its about more than just the network Measurement and Monitoring Is the link working? Who’s using it? Are they doing ok? Working with end users to support science! Better data access == Better science Who do you ask when you data transfer isn’t right? Rob Gardner: Collaborative Science begins with Data Transfers

Harper Reed, Obama’s second campaign CTO at IU Statewide October 2016

NetSage: Network measurement, analysis, and visualization http://portal.netsage.global

What does NetSage tell you? Today: How used are the links? Where are congestion points? Soon: What science domains are using the links? Identification of “problematic” flows

NetSage Data Sources Passive measurements Active Measurements SNMP Flow data from routers Packet header inspection from TSTAT Active Measurements PerfSONAR

NetSage Portal (2) http://portal.netsage.global

NetSage Portal (3) http://portal.netsage.global

NetSage Portal (4) http://portal.netsage.global

Year 3 Focus What are the top sites that use the IRNC links? What are the top science projects that use the IRNC links? How can we best identify a list of top talkers for each link?

“Why is understanding performance hard?” Why do scientists find it hard to move data effectively? Why is it hard to find the right network person to talk to about problems?

“Why is understanding performance hard?” Why do scientists find it hard to move data effectively? Why is it hard to find the right network person to talk to about problems? Ron Hutchins: There is something intrinsically hard in what we’re trying to do

Why is understanding data movement hard? Lots of pieces - Host system through networks to host system No one controls all the pieces Unknown expectations for what performance should be Theoretical vs pragmatic Asymmetric paths Soft failures are hard to find Many, many points of coordination

Building an army of self sufficient and diverse engagement practitioners around the world Help do this beyond DOE

Building our Engagement Army Joint Center being developed with ESNet Center for Engagement and Networks Provide support and assistance to anyone trying to transfer data Provide resources and assistance to help Application scientists Network and campus engineers People training up to be an engagement specialist

What do we do? Promoting effective technologies End to end performance DMZ, DTN PerfSONAR Data analysis End to end performance Holistic approach

What do we do? Promoting effective technologies DMZ, DTN PerfSONAR Data analysis End to end performance Holistic approach Jerry Sheehan: We don’t have a network architect – we have people who run networks

Two Types of Engagement Case Studies Walk through your science Understand needs and planning Think of this as regular maintenance, oil change, or planning to buy a new car “Roadside” Assistance This file transfer worked last week! Think of this like a flat tire, crash repair At IU this is done by both IN@IU and the IRNC GlobalNOC Performance Engagement Team

If we want to scale… Measurement, monitoring, and engagement are linked The people problem is as hard or harder than the technical ones Need an army of the willing More information: Jennifer Schopf jen@iu.edu

Acknowledgements IN@IU is funded by NSF awards #0962968, #0962973, #1450904, #1540933, #1638863 The TransPAC PacificWave 100gb/s network fabric is provided by Pacific Northwest GigaPop More information: jmschopf@iu.edu