Summary of Research Activities - October 2004

• Passive Measurement and Analysis (PMA) Project
• Active Measurement Project (AMP)
• Outreach, Collaborations, and Activities supporting Network Research
• Documentation, Web Work, Utilization Improvement, Publications
• Activities of each individual on the project

Passive Measurement and Analysis (PMA) Project

~ Continuing development of new metrics and real-time analysis for PMA

Klaus Degner is visiting us for four months to work on real-time approaches to passive measurements. He arrived early this month. Klaus D. is an undergraduate student at the University of Leipzig, Germany, where he works with our long-time research partner, Klaus Mochalski (who just spent two months in San Diego working with us). He will be further developing a real-time analysis application that he began there. This month, he made good progress in preparation for our demonstration of the tool at SC2004:

• Finished work on the program to convert traces into static html pages.
• Ported the tools from Linux to FreeBSD.  
• Programmed the distributed real-time analysis tool.
• Wrote the code for data analysis for the real-time application.
• Performed real-time monitoring of one SDSC link; tried to find out why the tcp-syn rate was so high on the link. Also did real-time measurements of the University of Leipzig uplink.
• Worked with the Endace install CDs, adapting the code to work with the new DAG 2.5.2 driver to support the DAG 6.2 cards. There were many problems to resolve, but porting the programs to DAG 2.5.2 was accomplished. Remaining problems between the DAG 2.4.x and 2.5.x driver series were removed; both of them can be used now.
• After installing the 10Gig Cards, tested the real-time tools on the server for SC2004 with a hardware packet generator.
• Improved the HTML pages for SC2004.
• Wrote extensive documentation, as well as some tutorials so others can integrate their analysis part very fast. Installed a Wiki to serve as a documentation platform.
• Early in the month, Klaus D. had a very long phone call with Klaus Mochalski. They both write code for the same library so they discussed coordinating our work and future plans for their library. As a result, he modularized their library more.
• Joerg and Klaus D. had a good technical planning meeting mid-month regarding development of the real-time software, database, and GUI architecture. Shortly after, they had a conference call with Klaus M., mainly talking about project planning longer term. Earlier in the month, the three had a conference call regarding Klaus D's current software development work.

~ Special Traces

We have documented the passive data collection from the HPWREN gateway at SDSC. Joerg and Matthew figured out that the data collected from the two points does not get sequenced as yet by the tshdump tool. Therefore, Joerg wrote an additional pipeline utility tshseq to order the different inputs via queuing in the order of the arrival timestamps. http://pma.nlanr.net/Special/hpwren.html.

Joerg had a long dialog with Internet2 and Caltech/CERN folks regarding capturing one of those bandwidth challenging megaflows running across the IPLS router instrumentation. One of the arguments that came up late is that folks are trying to create optimum conditions for the data transfer and not disturb the rest of the Abilene customers, whereas we would rather want to study the impact of such gigaflows on legacy traffic. I think we might engage into a series of tests here, eventually.

~ New (and developing) strategically important measurements and deployments

Development of passive monitoring for a lambda network (first stage prototyping of lambdaMON) ~

Joerg had a meeting at CISCO's optical division in Monza, Italy to discuss the situation regarding building lambdaMONs; the talks went very well. Afterward, he felt that we can expect support from the group as we move forward.

Before going to Italy to attend IMC2004 (etc.), Joerg came to San Diego. After completing preparations, he and Jim traveled to CENIC HQ in Cypress, CA, for testing of the first stage lambdaMON with the 15808-LH setup. They received tremendous support from Darrell Newcomb, Cenic Core Engineer. During the two days, they ran a wide range of tests, the net result of which, is that we now know:

1. The Tunable Channel Filter by Iolon is performing as expected in the field, it is capable of operating at low light levels (important for lambdaMON architecture) and delivers a good passband for 10GigE lambdas (we did test OC192c PoS, not 10GigE). What we could not test is channel isolation, however, we could see very clear clipping of any signal when we tuned off-channel, and we have no reason to believe the product would cause us any issues in field deployment.
2. The setup of collecting DWDM channels from the MON ports and amplifying them before feeding them back via receive transponders is working. We had an operational setup with two DWDM terminals and a few hundred kilometers of fiber, transmitting data without loss at OC192c PoS with rates between 10% and 100% using the Adtech AX4000 and the NLANR OC192MON collecting the very same data happily for hours. In total we peaked at 7.5 million packets/second and pumped a bit more than a Terabyte of data through, capturing every single bit without error.
3. The amplification is an ongoing headache, and we may need a pair of amplifiers (double amplification) as some of the signals in the field are even lower than in the lab and will cause intermittent results. This will come at a further cost increase.
4. Following a site survey at the CENIC POP downtown Los Angeles (818 Wilshire), we have a couple of new ideas for further collaboration with CENIC that we need to follow-up.

All in all, an extremely intense and productive two days. Joerg did get a chance to talk for a few minutes with David Reese (Chief Technology Officer, Cenic), Debbie Montano (Director, Development and Operations, NLR)as well as Tom West (CEO, NLR; formerly CENIC CEO), keeping them in the loop about our progress.
~~

Hans-Werner and Joerg met for a few hours when Joerg was in town, to develop some concepts for future PMA activities. The meeting went very well, as good progress was made.

~ Upgrades, troubleshooting, and maintenance on the PMA servers and infrastructure

We are working with Mike Gleicher to enable the HSI interface to the HPSS, so we can migrate away from ftp access.

Very busy month with preparations for multiple back to back events, including the lambdaMON trials at CENIC (in order to have things working before SC2004). Also, we are trying to get an OC192MON into the SC2004 showground in time. And, we are trying to get all the remaining gear for the IPLS router clamp into Indianapolis before November 12, so Joerg can complete the work there as a sidetrip after SC2004 in Pittsburgh. The gear for the lambdaMON, the 10Gb monitor for SC04 and installation at IPLS and the additional OC3/ OC12 monitor machines were all ordered and received.

Unfortunately Endace shipped three DAG6.2 and one DAG6.1; we are trying to match our original requirement for 2 and 2 and have had a positive response from Jason Hurd to have this organized shortly. The 10Gig Cards were installed in the server for SC2004. There were some problems with the new DAG driver and one card was broken (replaced for time being). Endace will be sending us the needed DAG6 cards for SC2004.

Joerg gained an extra day in his schedule (while in SD) so he and Jim used that time to upgrade the PMA data server to FreeBSD 5.3-RC1, with a few hiccups, but all working. They will watch the performance and see if it makes a difference to the stability.

Support and troubleshooting, existing PMA measurement sites:

Worked with several people at Ohio State University on getting the passive monitor there functioning properly again. It took a week and several emails for the site to find our machine. They found the monitor and cycled it for us and it is again collecting normally.

Joerg is planning on traveling to fix the machines at NCAR and Front Range GigaPOP, after SC2004.

Active Measurement Project (AMP)

~ Progress on the reimplementation of AMP and the development of a new testing architecture

• Quite a bit of progress on both the AMPlet and AMP central code in October:
• One of the AMP meshes running the new code had some odd results because two instances of the software accidentally got started at the same time. This hurried along adding a PID file to the three daemons (measured, xfer and xferd). At startup, the file is checked and if the process identified there is still active, the new one refuses to start. Override instructions are included in the error message. We also added daemonization and the infrastructure for an (ssl encrypted) control interface to the three daemons. The control interface can currently connect to a local or remote AMP process and execute some test commands and give help.
• Finished the packaging of the AMP central code which turned out to be a bit more complicated than expected, with quite a few things hard to locate on an arbitrary system, (for example, the URL that corresponds to a particular directory). Worked through several installs of the AMP central code and ironed out a few missing bits etc.
• Added a new graph type to the displays, a CDF of loss against packet size. It was fairly easy to do and add in, which is good confirmation that we have the basic structure of the code right.
• Added another call to the AMP extension for getting the current time in any time zone. This is useful to know what days to display on the Web pages.
• We began writing code that will terminate an external test if the external test program happens to lock-up/fail (external test timeout). Continuing at the end of October.
• The ip-hopcount graph was added to the AMP central code; additional slight modifications were also made. This included some small changes to dgraph to ease its use. The monitor map graphic was also added.
• Added dns-names to the traceroute graph. For the moment we are simply looking up the current dns value and doing some very basic caching of the data. We also performed some work with the traceroute graphs in relation to memory issues.
• Wrote a script to make updating of the database much more automatic when changes occur. This allows us to change the values in the install configuration and then have that change happen once make install is run.
• The AMPlet code is largely finished. Work continued on getting the code to do a buildall cleanly, and all but one of the tests are now repeatedly passing. The one which is not working only appears on certain machines.
• Work continued on getting the AMPlet code to compile and pass all make check tests on an older Debian machine.
• We discovered a problem with the cvs checkout process; it did not run the check out scripts when a remote checkout was done. This was sensible, but meant that many of the tests were not working. Therefore the script was moved to the install process, instead.
• Began evaluating adding the RTT distribution CDF graph.
• Good progress was made on the IPMP code in the central code; it is close to finished.
• Work was done to bring the IPMP IPv6 implementation into line; then the code was handed over for use in testing the new userland code.
• Started work on the diurnal variation aspect of the code. We are trying to determine how to recognize diurnal variation on a link so that it will not be treated as an abnormal spike in performance. This has met with limited success working with the median and averages. Next we will look at using stddev to create a range at which data will be considered acceptable.
• The amp central install was fixed so that it works on Debian sarge. We setup one of our test systems under sarge to do the testing. We also did so on CRCnet's peon server (which has virtual hosts and TCP socket based authentication for postgresql).
• This Debian sarge fix and testing was good for improving the generalizability of the package. However, Tony is increasingly coming to the opinion that, it will be impossible to make the package install smoothly in every environment. Instead, the hope is to provide good documentation regarding what to do when something goes wrong.
• A perl script was written to clean up the doubling problem which caused double entries in the data.
• Battled with two bugs that refused to be solved easily. Both occurred only in FreeBSD testing. The first turned out to be a 32/64 bit problem based around the off_t type (which is 32 bits on most systems but 64 bits on FreeBSD). The second turned out to be a bug in the Linux NAT that we use. It corrupts the checksum on and ICMP packet that is all zeros. Both took ages to find and were very frustrating.
• While working on the central code which calls dgraf, took care of small bugs/feature additions.
• Discovered a bug in the tracetest program; came up with a decent, although not ideal, solution. Also discovered that there was probably a memory leak in the php code, which we think is now corrected.
• We gave some support to Matt Brown (of the CRCnet team) who is writing some raw data Web pages. (He wants to give datasets to some NZ statisticians.) Matt completed a set of raw data scripts and Web pages.
• Warren Matthews was in contact about a demo of the Webservices interface that he is working on. We filled him in on the AMP extension and assisted him with some other information and some suggestions for debugging of code with which he was having problems.
• A plan for doing updates to the databases when there are new releases was developed.

~ New (and developing) strategically important measurements and deployments

Preparation and shipment of a new AMP machine for the University of Washington, C & C department.

Continued work to prepare the machines destined for the CENIC network. Darrell Newcombe is determining which of the locations will be power sourced from a -48 VDC power bus.

For the planned deployment TBD in Russia, the current plan is to have Greg Cole hand-carry an AMP on his next trip to Moscow. Later he will give us an updated list of sites in Russia (maybe also some sites in China) for the one-way measurements from the AMP on the NaukaNet link at Starlight.

10GigE AMP ~ Testing continued with the 10GigE machines, with the max speed now close to 5 Gbps.

Other AMP meshes:

Jose Monterio from UNIFACS in Brazil has deployed three local AMPs for evaluation, which if successful, will go nationwide.

Another site joined the NZ Amp mesh (an ISP called Catalyst) making seven in all, mostly at ISPs. We had a few problems setting it up, mostly because of "human error."

Tony attended a meeting in Auckland with Telecom (NZ's largest telecommunications company). They are doing a fiber to the home trial (to ~1000 homes) and are looking for research partners. He talked to them about ways we might support their development lab and was hoping to interest them in deploying an AMP infrastructure. They are possibly interested in using AMP or passive measurement. It is not clear whether it will go anywhere at this stage. However, they are using AMP for another project, so there is potential.

~ IPMP

A post to the IMRG mailing list indicates there is actual demand for the IPMP work.  http://www1.ietf.org/mail-archive/web/imrg/current/msg01145.html

~ IPv6 and IPv6 Scamper

Good progress continued on Scamper:

• Wrote code to make Scamper operate in a daemon mode, to complement the traceroute-like and with-list modes of operation. Fixed up the logic a bit in preparation to make it behave as a daemon that is controlled by commands sent on a socket connection.
• Enabled the control socket code. At this time, one can start Scamper in daemon mode and then login to the process with telnet, and setup the default parameters. It can receive and act on addresses to traceroute.
• Working on finishing up the code to specify 'sources' in the control code. The code is already there and working in the command line version of IPMP. All of this code is non-blocking, so read/writes from control sockets will be buffered and not cause Scamper to block until the system is ready to take the data.
• Fixed up the main-loop so that the timeout it passes to kevent/select reflects how long it will be until it actually needs to send another probe. The code to do these tasks was non-trivial to add, as it is fundamental to the operation of Scamper.
• Reworked the address list code so that Scamper does not block when reading a large address list. It now reads the file as the OS is ready to provide chunks without blocking. Could not get kqueue() to signal end-of-file so went back to select() (for now). Also wrote code so that Scamper will merge many lists together, each with a different priority.
• Finished the process of privilege separation. Privilege separation involves forking a setuid root process and dropping privileges permanently in one of the two processes. The 'lame' process does most of the work in a chroot environment, while the privileged process does things that require privileges. For Scamper, this means opening BPF devices, deleting cloned routes on FreeBSD and MacOSX where necessary, and opening input/output files as the underlying user. The idea with privilege separation is to make it much more difficult to obtain root through any flaw in the unprivileged code, and to contain any damage possible by compromising the lame process. Also took the time to update the route socket code, so that Scamper does not block waiting for a response.

~ Upgrades, troubleshooting, and maintenance on the AMP servers and infrastructure

Testing and transition to the new AMP servers (AMP2 and VOLT2) ~

• The beginning of October we brought the AMP server back online collecting data along with VOLT. Several unforeseen anomalies cropped up to slow that process but it is now up and running. That was done to provide for a change in direction on the new AMP2 and VOLT2 servers. Both of the old servers, AMP and VOLT, are functioning well and reliably performing the data collection and Web page serving for the AMP data. However, we are working on what appears to be a sshd incompatibility with AMP and VOLT.
• We researched and installed the latest Debian release with the 2.6.8-1 kernel on the new VOLT2 server. This proved to be more difficult than anticipated. Particularly troublesome was the driver for the 3Ware RAID controller configured to run RAID10. In this later release of Debian, designators have been changed, including the 3Ware driver. Locating the correct designator solved the problem. Another issue was the Intel GigE NIC's and the peculiarities in the /etc/network/interfaces file for the 2.6.8-1 kernel. Previous to this, the additional disks needed to implement the RAID10 scheme and utilize a separate system disk were installed in the Volt2 server.
• The VOLT2 server, with the RAID10 disk setup, is installed in the NLANR rack, in the SDSC machine room, connected to the .123 network and the backup network (10.28.5.x). It was set-up using the /etc/hosts.allow and /etc/hosts.deny files. One of the RAID10 disks suffered a failure and was replaced. Except for some security questions, which we are continuing to be improved, it is ready for the data collection and serving software.
• The additional disks needed to implement the RAID10 scheme will be added to the other new server (AMP2) and the FreeBSD5.3 operating system installed. This will enable the comparison of the Debian Linux platform to the FreeBSD platform running the AMP data collection software.
• Early in October, archiving of the AMP server was needed just before the normal maintenance period on HPSS. However to prevent the archive process from halting, we modified the doarchive script on AMP so it would continue to try to connect to HPSS for eight hours. That worked well and the archive continued until the data disk fill on AMP was down to the high seventy percent range.

Late in October, the system disk on the photon.nlanr.net system manager machine crashed. We conferred with Todd Hansen; unfortunately, there is inadequate backup for this machine. We tried to resurrect the machine with the processes that it runs. At this time, it appears that the most difficult process will be to restore the private ssh identify file. We anticipate that manual initialization of AMPlets in the field may be necessary until the system is completely restored.

SOAP was set-up on the volt server for Warren Matthews.

This month we observed international sites apparently blocking ICMP traffic from amp-kiwi (U. of Waikato). Our investigation found that these sites block all traffic from commodity networks. And, of course, all networks in New Zealand are commodity. We are working with each of these sites to have them change the filter to permit ICMP packets from the amp-kiwi site.

Support and troubleshooting, existing AMP measurement sites:

Site outage remains at a very low level. A total of 12 remote sites in the AMP Network meshes received attention during this period; most were resolved and the monitors are again collecting data. Only four were still being investigated, or pending site action, at the end of the period. (Outages are considered "open" until the monitor is again collecting data.)

12 problem sites:  8 resolved, 4 open - at the end of the period.

amp-cnic (Computer Network and Information Center in Beijing, China) ~ experienced some brief outages this period. Each one seems to have been related to a continuing unreliable power source. The site technician is looking into options to improve the power reliability. Apparently, there are some improvements planned for powering that part of the facility.

amp-jpl (Jet Propulsion Lab) ~ had a short outage due to a mishap with power source changes; corrected shortly after site technicians were consulted.

amp-korea (Korea KISTI site) ~ a replacement AMP monitor was prepared and shipped, it is currently in Korean Customs. We are working with the site technician, Wanming Lee, to get it released and hope to have the site back online shortly. This is our only continuing outage; the site has been down for a few weeks.

amp-mit (Mass. Inst. of Tech.) ~ had short outage due to a minor error in network connect changes; corrected shortly after site technicians were consulted.

amp-naukanetnwu (NAUKANet at Northwestern University, Chicago) ~ was down. Greg Cole informed us that the machine is being moved to another point on the network; we expect the machine to be back online shortly.

amp-ncren (North Carolina Research and Ed. Network GigaPop) ~ was down for short time. The cause was a power breaker which had been inadvertently switched off, shutting down the rack with the AMP monitor and some additional servers. Problem was corrected shortly after we asked the site to investigate the loss of connectivity.

amp-orst (Oregon State U.) ~ went down this month. We requested that the site technician investigate the problem. He found that the machine was powering up, however it was producing no video or network activity. At his request we prepared a replacement machine and it is ready for shipment.

amp-ou (U. of Oklahoma) ~ is back online after a couple of week outage. This was due to a site network blockage issue that was finally resolved.

amp-psc (Pittsburgh Supercomputer Center) ~ is down; site people have been asked to investigate.

amp-surf (SURFnet in Amsterdam) ~ again had a network interface problem that we corrected remotely.

amp-uab (U. of Alabama, Birmingham) ~ site outage was due to a faulty ethernet cable. The connection was corrected when the cable was replaced.

amp-unin site (UniNet in Thailand) ~ was down for a short time this period. Site people restored it after our inquiry. The correction, as reported, was merely re-plugging the ethernet cable into the switch port.

Outreach, Collaborations, and Activities supporting Network Research

~ Papers, Presentations, and Conference/Meeting Participation

Joerg attended IMC2004, Internet Measurement Conference 2004, October 25-27, 2004, Taormina, Sicily, Italy. He found it an excellent opportunity to be in dialog with a broad range of people and gather ideas for collaborative work.

Late in the month Klaus D. shared the latest version of his real-time application tool which will be run live on a monitor at SC2004 in Pittsburg. Joerg is quite pleased with the results "... and it looks like we are all set. Good work, Klaus!"

Plans were made for Tony to give a presentation on the new AMP software release at the February Joint Techs meeting.

Slides were made on the plans for deployment of a local AMP mesh in Brazil. One summary slide and another which displayed a map with the three locations under evaluation.

Joerg submitted a four page paper/abstract on lambdaMONs for consideration to PAM2005.

Tony is a coauthor of two submissions to PAM2005, a third paper was not submitted.

~ Collaborations And Activities Supporting Network Research

Chip Cox and others, AMPATH ~
Ronn and Tony had a conference call with them about passive and active measurements on links to Mexico and Brazil.

Darrell Newcomb ~ Cenic
Joerg and Jim received tremendous support from him while testing the lambdaMON equipment in Cypress at the Cenic HQ.

Greg Cole (Gloriad) ~
Joerg and Ronn had an excellent conference call with him to understand more about his problems and needs for monitoring. We are intending to follow up with a meeting somewhere and sometime during November, if it can be arranged.

Human Computer Interface group, Univ. of Waikato ~
Tony built the basics of an nxn raw data page for them, as they are interested in doing new visualizations of active and passive datasets.

Matt Zekauskas, Internet2 ~
At IMC2004, Joerg had an intensive discussion with him regarding SC2004 measurements and the IPLS router clamp. Also worked with him and Rick Summerhill to determine timing and equipment needs for IPLS.

Paul Love, Internet2 ~
Ronn followed up with him regarding an AMP presentation by Tony at the next Joint Techs meeting.

Debra Montano and Tom West, National LambdaRail Inc (NLR) and David Reese, Cenic ~
While testing in Cypress, Joerg talked to them to keep them in the loop regarding progress with the lambdaMON prototyping.

Aaron Chen, OptiPuter ~
Ronn met with him and they had a conference call with Tony regarding active measurements for the OptiPuter project. Aaron will check with others on the network information that would be helpful and get back to us.

Jay Dombrowski, SDSC NetOPs ~
While in SD, Joerg had a chance to talk to him, discussing the possible use of TeraGrid rack space at the LA POP and they drifted away into how we can help with measurements on solving their 40 Gigabit transmission problems between Chicago and San Diego.

Marek Hajduczenia, SIEMENS in Lisboa ~
Joerg helped him resolve problems downloading PMA data onto his Microsoft platform.

Jose Monterio, UNIFACS, Brazil ~
Ronn and Tony had a conference call with him regarding active measurements in Brazil. He has deployed three local AMPs for evaluation, which if successful, will go nationwide.

Nikolas Kazepis, UOC (Institute of Computer Science, FORTH), Greece ~
Joerg is helping him with Code Red traces from July/August 2001. Once retrieval from the HPSS is complete, Joerg intending to publish these as a separate data set as well.

Joerg had email conversations and phone calls regarding short-term plans (October and November) and future passive measurement possibilities with a very large group of people, including David Reese (CTO) and Brian Court (Operations) at CENIC, Jay Dombrowski (ENS), Anja Feldmann (TU Munich), Klaus Mochalski (Uni Leipzig), Gary Betteridge (Waikato), Malathi Veeraraghavan (Uni Virginia), Mark Johnson (NCREN), Bill Wing (ORNL), Balachander Krishnamurthy (AT&T Research), David Hamilton (Waikato), Peter Arzberger (SDSC), Vijay Samalam, Kevin Thompson (NSF), Ronn, Tony and Hans-Werner.

Documentation, Web Work, Utilization Improvement, Publications

A poster was created on our efforts to prototype passive monitoring for DWDM optical networks, such as the National LambdaRail (NLR) - lambdaMON. Roll up versions were created for use at SC2004 and similar venues. 25 of the smaller size were printed for distribution at, and sent with the two roll-up posters to, SC2004. Feedback was received from most partners (NLR, Internet2, CENIC) on the final draft.

Two additional NLANR/MNA published papers Web pages were created to make it possible to differentiate the published papers by whether they are PMA or AMP. There are now AMP- and PMA- only pages. Each links to the others, and the page with all papers is still the primary interface (as the index file), with easy access at the top to the others. http://moat.nlanr.net/Papers/

Prepared for SC2004, by printing copies of the AMP PathVis handout and arranging with Kinko's to print 100 copies of the recent Network Analysis Times. They were sent with the SDSC crates to be trucked to Pittsburgh (along with the needed machines and equipment).

PMA Collection and Use Statistics (Web logs) project ~ Nearing completion of the Web logs project. The system in place seems to maintain itself well and there only a few modifications and/or additions left, primarily to the graphs and structure of the HTML pages. We need to change the units/scaling on some graphs and create a generic index page for the stats directory so that each part of the stats pages is directly accessible from there.

We fixed a small bug in the (old) AMP database to Web code that stopped some updates to site data from making it to the Web pages.

We had some Web server problems with the nlanr.net/moat/mna names, which was due to the upgrading of moat quite some time ago; it was fixed.

Activities of each individual on the project

AMP team

• Tony McGregor ~
  AMP reimplementation, central and AMPlet code; working with Jeremy new AMP servers transition issues; collabs.
• Jeremy Kallstrom ~
  AMP reimplementation (external test timeout, monitor graphic, traceroute graphs, IPMP, bugs, etc.); testing the 10GigE interfaces.
• Matthew Luckie ~
  AMP IPv6/Scamper; AMP IPMP; help with HPWREN tshdump script.
• Bud Hale, Jim Hale ~
  photon crash; new machine prep and deployment; new AMP servers transition reconfig/testing; working with sites, troubleshooting existing infrastructure.

PMA team

• Jörg Micheel ~
  Cypress lambdaMON prototype testing; prep for SC2004 measurements and IPLS deployment; came to SD; lambdaMON poster (text, etc.); collabs; IMC2004; HPSS/HSI; HPWREN page; PMA server upgrade.
• Chris Gross ~
  PMA Collection and Use Statistics (Web logs) project
• Klaus Degner ~
  Real-time application software development, Web pages, etc.; 10Gig card installation; help with SC2004 and IPLS prep (installations, etc.)
• Jim Hale, Bud Hale ~
  Cypress prototype testing; preparations for SC2004 and IPLS instrumentation; troubleshooting existing sites; PMA server upgrade.

MNA, AMP, and PMA Outreach, Documentation, Web work

• Maureen Curran ~
  Write/edit (September monthly report; 3Q 2004 quarterly; lambdaMON poster [ed.]); handouts and NATimes for SC2004; new Web pages for published papers; Web logs project pages.
• Gina Intriligator ~
  lambdaMON poster.

Management and Administrative

Hans-Werner Braun, Ronn Ritke, Tony McGregor, Jörg Micheel ~
Weekly NLANR/MNA managers conference calls.

• Hans-Werner Braun ~
  Met with Joerg re PMA direction; HPWREN measurements; moat/mna server problems; etc.
• Ronn Ritke ~
  Met with Brazilian network engineers; collabs (multiple conference calls); lambdaMON poster; meetings with Joerg; SC2004 prep; various meetings with staff attended SDSC reorganization meeting; worked with Mary re new online budget process; budget re lambdaMON equipment; paperwork for Klaus D.;

- 30 -

Top       2004 Oct 18       NLANR/MNA home page