1) Scientific research increasingly relies on large-scale data transfers between collaborating institutions over high-speed networks. 2) ESNet provides high-bandwidth connectivity between DOE sites but needs to efficiently allocate guaranteed bandwidth for transfers. 3) The presentation proposes enhancements to ESNet's OSCARS reservation system to suggest optimal reservations that meet researchers' bandwidth and timing requirements for data transfers.

1. Advance Network Reservation
and Provisioning for Science
 Mehmet Balman
Scientific Data Management Research Group
(with Arie Shoshani and Alex Sim)
Lawrence Berkeley National Lab
Summer Presentations - Aug 13, 2009

2. Outline
 Introduction
 Resource Management
 ESNet and OSCARS
 Network Reservation
 Problem, Solution, Methodology

3. Introduction
 We are in a new era that offers new oppurtunities to
conduct scientific research with the help of
computation
 Computational intensive science: particle physics,
climate modelling, bio-informatics simulations
 Scientific simulations and experimental facilities
generate massive data sets
 Climate modelling data
 35 terabytes shared by more then 2500 users worldwide,
 Next generation archive will be more than 650 terabytes
 Large Hadron Collider
 Expected to generate 100gigabits per second

4. Introduction
 Large scale application necessitate collaborations
 Data need to be tranferred to remote sites for
further analysis (validate with simulations)
 Need on demand high speed data access between
collaborating parties
 High performance visualization
 Large volume data analysis
 Require mass storage systems
 Need coordination and management of resources
( BeStMan: Berkeley Storage Manager)

5. ESNet (Energy Sciences Network)
 Provides high bandwidth network interconnect
between more than 40 sites
 Connecting experimental facilities,
supercomputing centers and thousands DOE
scientists
 Delivering network as a service
 Predictable performance
 Efficient resource utilization

6. OSCARS
 The ESNet On-Demand Secure Circuits and Advance
Reservation System (OSCARS)
 Conducts a QoS path for guaranteed bandwidth
 End-to-end provisioning between multiple domains
 Guaranteed bandwidth (at certain time, for a certain
bandwidth and length of time)
 OSCARS components include reservation
manager, Bandwidth scheduler, and path setup
system
 Needs to have information about current and future
states of the network

7. Data Movement
 End-to-end High Performance Data Movement
 Network reservation
 Provisioning in client sites
 Storage Allocation
 Therefore, we need coordination between Storage
Resource Managers and Network Resource
Allocation

8. Network Reservation
 Users make reservation over a web service
interface
 Reservation request:
 source/destination end-points
 Requested bandwidth
 start/end times

 What is requested bandwidth can not be
guarantted?
 Try-and-error until get an available reservation

9. Reservation request
 Not aware of possible options
 Cant make an optimal choice
 Ineffective use of overall system
 How can we enhance the OSCARS reservation
system?
 Submit constraints and the system suggests
possible reservations satisfying requirements

10. A new service
 Source / destination end-points
 Maximum bandwidth that can be used
 Amount of data requested to be transferred
(Volume)
 Earliest start time
 Latest completion time
 Criteria (reserver a path for earliest completion,
reserve a path shortest transfer duration)

11. Challange
 Dynamic max-bandwidth problem!
 Time dependent dynamic network
 The bandwidth value for every link is time
dependent
 Discrete time algorithms: time is modeled as a
set of discrete values and a static graph is
constructed for every time interval.

12. Example
 A vehicle travelling from city A to city B
 There are multiple cities between A and B
connected with separate highways.
 Each highway has a specific speed limit
(maximum bandwidth)
 But we need to reduce our speed if there is
high traffic load on the road
 We know the load on each highway for every
time period (reservations)

13. Example
 The first question is which path the vehicle
should follow in order to reach city B from city A
as early as possible?
 Or, we can delay our journey and start later if
the total travel time would be reduced. Thus,
the second question is to find the route along
with the starting time for shortest travel
duration.

14. Example
 But, we are dealing with bandwidth reservation
where allocation should be set in advance
when a request is received.
 We have to set the speed limit before starting
and cannot change that during the journey.
 Advance Bandwitdth Reservation

15. A new algorithm
 Search interval is divided into time windows
 A time window represents a period of time
where we have a stable status of available
bandwidth of all related links
 A snaphots of the network topology
 The algorithm should be fast and scalable. Presenting
clients/users possible reservations requests and
alternate options

16. What I did?
 Developed a new approach
 Analyze the algorithm and tested with large
graphs
 Implemented a library to be integrated into
OSCARS
 If interested, a technical talk with a short demo
on Monday at 2pm in 50F-1647

17. Special Thanks to
 Arie Shoshani
 Alex Sim
 Evangelos Chaniotakis
 David Robertson
 Mary Thompson
 ESNet team
 Scientific Data Management Research Group