Core Measurement Infrastructure

The core measurement infrastructure for LHC-MI is intended to reliably provide a number of network and host performance parameters along end-to-end paths.
This will extend and leverage existing and emerging measurement projects to meet the needs of the LHC-MI to provide both network-centric and cluster-centric performance data.

We expect to gather and use data from the following network sources to provide an end-to-end performance data resource:

The IEPM-BW toolkit, developed at SLAC, is currently deployed at monitoring hosts at about 40 sites around the world, including major measurement hosts at SLAC, CERN, FNAL, BNL and Caltech. 
These measurement hosts run active end-to-end light-weight measurement tools, such as ping, traceroute, pathchirp [PATHCHIRP] and pathload [PATHLOAD], and heavy-weight measurement tools, such as thrulay [THRULAY], iperf [IPERF], and GridFTP [GRIDFTP] at regular intervals.
The light-weight, more frequent measurements, are used to assist in interpolating the less frequent, more heavy-weight measurements in order to reduce the strain on network resources.
The type of data collected by these measurement tools includes round-trip-time, hop-by-hop router response, capacity and available bandwidth, achievable throughput and file transfer rates.

A second component of the monitoring infrastructure will be from core routers and switches which form the backbone of the Academic and Energy Science Networks around the USA (and in Europe). 
The Abilene Measurement Infrastructure (AMI) [AMI] and the perfSONAR [PERFSONAR] projects will provide router interface utilization and capacity data using standardized schemas and web service facilities.
Both projects allow data access using standardised web services interfaces.
AMI is currently deployed across Internet2's Abliene network and passively measures 11 core routers.
There is currently much momentum behind the perfSONAR project with support from Energy Science Network (ESnet) and the Pan-European Geant network and Internet2 to provide uniform data access to all routers with much of ESnet's and Geant's router interfaces are available through perfSONAR.

A third source of measured data from existing networks is Netflow [NETFLOW] passive measurement data that can be obtained from select routers, in particular, border routers at collaborating sites.
The Netflow records from a given router will be collected by a host collocated in the AS of the router.
This host will suitably anonymize and select relevant records (e.g. long lived flows, selected ports/applications, etc.) and make them available.
A goal of these measurements is to obtain data on the start times, transfer sizes, end times and characterizations (e.g. top talkers, transfer rates, arrival rates and call holding times) of the traffic coming in and our of the AS.
As a crude model, we propose using flow data to gather statistics on long-lived flows for file transfer applications and end site destinations to provide basic account facilities.

We expect to collect host data from the following sources [Yee: i really don't know much about this stuff]:

Ganglia... etc.






[PATHLOAD] M. Jain and C. Dovrolis. Pathload: A measurement tool for end-to-end available bandwidth. In Passive and Active Measurement (PAM) Worksop, 2002. 

[PATHCHIRP] V. Ribeiro, R. Riedi, R. Baraniuk, J. Navratil, and L. Cottrell. pathChirp: Efficient available bandwidth estimation for network paths. In Passive and Active Measurement (PAM) Workshop, 2003. 

[THRULAY] S. Shalunov. thrulay, network capacity tester. unpublished, available at http://www.internet2.edu/ shalunov/thrulay/. 

[NETFLOW]  Cisco Systems. Cisco netflow. http://www.cisco.com/warp/public/732/ Tech/Netflow. 

[IPERF] A. Tirumala, F. Qin, J. Dugan, J. Ferguson, and K. Gibbs. iperf version 2.0.2. unpublished, available at http://dast.nlanr.net/Projects/Iperf/.

[GRIDFTP] B. Allcock, J. Bresnahan, R. Kettimuthu, M. Link, C. Dumitrescu, I. Raicu, and I. Foster. The globus striped gridFTP framework and server. Submitted to the 2005 High Performance Distributed Computing Conference (HPDC 14). 

[PERFSONAR] Hanemann, A., Boote, J. W., Boyd, E. L., Durand, J., Kudarimoti, L., Lapacz, R., Swany, D. M., Zurawski, J., Trocha, S., "PerfSONAR: A Service Oriented Architecture for MultiĞDomain Network Monitoring", In "Proceedings of the Third International Conference on Service Oriented Computing", Springer Verlag, LNCS 3826, pp. 241Ğ254, ACM Sigsoft and Sigweb, Amsterdam, The Netherlands, December, 2005.

[AMI] W. Matthews, Abilene Measurement Infrastructure, unpublished, available at http://abilene.internet2.edu/ami/webservices.html