The annual Computing Picnic was a success! Beautiful weather and a great turn out made for a fun event.
Check out photos from the picnic here.
Starting Oct. 1, CCD will be responsible for the Duplicating and Printing Services and the Audio Visual Services for the Directorate conference rooms and Cindy Arnold joins Computing.
Mark your calendar! Computing will host a Computing Technology Day in Wilson Hall on Tuesday, Oct. 6, from 9 a.m. to 1 p.m. An industry exhibit featuring 20 companies will be on display in the atrium along technical presentations in One West. No registration is required.
Eileen Berman will transfer to SCD on Oct. 1 as her LBNF/DUNE role has expanded. Jerry Guglielmo will become the new Service Operations Support department head.
Have you seen the snazzy new digital sign in Wilson Hall? Make sure to check out the hard work of Laura Mengel, Randy Reitz, Scott Nolan and John Galvan with content provided by the Office of Communication next time you're walking through the first floor of Wilson Hall.
Wilson Street gate access hours have been expanded. The gate will be open from 6 a.m. to 4:30 p.m. on weekdays. Outside of these hours, employees will be able to use their badge to enter at any time. Anyone can exit through these gates at any time. Click here for more information.
Irene Shiu (OCIO/Service Management/Business Analyst)
Employee Performance Recognition Awards
Eileen Berman: for implementing Fermi Managed Printing Services
Briant Lawson: for implementing Virtual Server Service
Jason Morris: for implementing Virtual Server Service
Michael Rosier: for implementing Virtual Server Service
Glenn Cooper: for his contributions to the 2015 art/LArSoft course
Patrick Gartung: for his contribution to the 2015 art/LArSoft course
Chris Green: for his contributions to the 2015 art/LArSoft course
Chris Jones: for his contributions to the 2015 art/LArSoft course
Kyle Knoepfel: for his contributions to the 2015 art/LArSoft course
Jim Kowalkowski: for his contributions to the 2015 art/LArSoft course
Marc Paterno: for his contributions to the 2015 art/LArSoft course
Paul Russo: for his contributions to the 2015 art/LArSoft course
Saba Sehrish: for her contributions to the 2015 art/LArSoft course
Erica Snider: for her contributions to the 2015 art/LArSoft course
Sheila Cisko (Computing Service Specialist III)
Briant Lawson (System Admin Manager)
Peter Rzeminski (Computing Services Specialist IV)
Olga Terlyga (Computing Service Specialist III)
Scott Nolan (Senior Enterprise Architecture)
Jim Amundson (Computational Physics Developer V)
Ken Herner (App Dev and Sys Analyst II)
Mike Kirby (Scientist I)
Andrew Norman (Scientist I)
Tim Skirvin (Computing Services Specialist III)
Ted Zmuda (Engineer III)
(5, 10, 15 & 20+ years)
John Marriner- 37 years
Gregory Cisko- 36 years
Luann O'Boyle- 35 years
Gregory Deuerling- 33 years
Gustavo Cancelo- 27 years
David Fagan- 27 years
Norman Ho- 27 years
Robert Harris- 26 years
Al Lilianstrom- 25 years
Patricia Cameron- 24 years
Stephen Kent- 24 years
James Annis- 22 years
James Simone- 22 years
Jon Bakken- 21 years
Yujun Wu- 15 years
Kyle Knoepfel- 5 years
Tyler Parsons- 5 years
Looking ahead: Super Computing Conference 2015
The 27th annual International Super Computing Conference for High Performance Computing, Networking, Storage and Analysis will be held in Austin, Texas from Nov. 15 to Nov. 20. Personnel from 15 DOE Office of Science funded National Labs will participate and contribute to the DOE booth on the conference show floor and in the various technical programs scheduled throughout the week.
Under the SC15 technical program, Fermilab has submitted three research posters and two workshop papers.
At the DOE booth, Fermilab will be conducting two demonstrations titled "HEP analysis framework" and "Network data analysis". Digital content from various labs will be displayed as interactive e-posters on 6 monitors. Fermilab will be highlighting 6 key projects, the primary being HEP Cloud. The featured DOE booth Fermilab talk will be on the HEP Cloud project and will be presented by Panagiotis Spentzouris.
At the SCinet NRE (Network Research Exhibition) Fermilab will conduct one demo.
Various Fermilab personnel will present at vendor booths on topics related to vendor products. Below is a list of vendors where such presentations are either already scheduled or have not yet been decided.
- KISTI - Korea Institute of Science and Technology Information (scheduled)
- Micron Technology Inc. (scheduled)
- Ciena (scheduled)
- Mellanox (TBD)
- NVIDIA (TBD)
- Intel (TBD)
The topics being presented at the vendor booths are as follows:
- HEP Cloud
- HEP Pattern recognition with an Automata Processor
- Application-oriented network traffic analysis based on GPUs
- WireCAP – software-based packet capture engine for commodity NICs
On August 16, SCD’s Bonnie King left on a short vacation to Seattle, Washington. However, this wasn’t your typical touristy vacation. Rather than souvenirs, King brought back knowledge. She attended LinuxCon, a three-day conference for people who interact with Linux in a variety of capacities to learn from each other and the leaders in the Linux community.
After attending last year’s LinuxCon in Chicago, King decided it was worth the trip to the West coast. “I’m a Linux administrator, so a lot of the content was applicable to my job. These conferences are a great opportunity for folks in Computing because they allow us to learn about new technologies and see what is happening in industry.”
Another draw was being able to experience in-person how a large-scale, open-source community like the Linux kernel functions. Attending such conferences, however, reinforces the fact that in the Linux world, she is somewhat unique. “Since I learned Linux online, at first I wasn’t confronted with the reality that this is a pretty homogenous group,” she said. “There are not many women, and there is not much diversity in general.”
King received free admission to the conference through the Linux Foundation’s diversity scholarship program, which aims to make conference attendance more diverse. The foundation features a profile about her on its blog.
Surprisingly, King’s avid love of Linux began as a hobby in college. As a drawing and painting major (check out her submission to the Fermilab employee art show in the Wilson Hall gallery through Nov. 30), she spent her free time exploring her interest in electronics and hacking. She turned her hobby into a career and has now spent four years handling technical challenges she finds exciting and taking advantage of the opportunity to make meaningful contributions here at Fermilab.
From the CIO: Progress toward a Computing strategy
Chief Information Officer Rob Roser
At our most recent All-hands meeting, one of the topics I discussed was the development of a computing strategy and an associated document. Here’s an update for you and a little more detail.
Why are we doing this? The idea behind this computing strategy document is to articulate for us and to the laboratory our vision for computing at the lab over the next five years. For the lab to be successful, both SCD and CCD must also be successful. Thus, we want this strategy document to create an “end-state” that describes where we want to be as an organization in 2020 and then articulate the main themes and priorities that must occur in order to achieve this vision. It should answer the questions of how we want to evolve and adopt or leverage emerging technologies and what our focus will be. We are obviously not big enough or sufficiently well-funded to be all things to all people. We must make important choices!
Strategy documents are not unique to this purpose. We have been writing them for years. However, from my perspective, there are two significant differences in this exercise than from those of the past. First, we are looking out further―five years. Second, we are engaging the leadership of our organization in the process rather than the traditional top-down approach in which only the most senior members participated. I feel that this plan has a much greater chance for success since the organization is playing a significant role in developing it.
To date, we had three half-day meetings engaging broadly the members of CCD, SCD and OCIO. We then broke up into four “theme teams” to focus on the following areas: workforce and organization, scientific computing, IT (core computing) and process and information security. After an initial set of conversations, the teams reconvened, and 10 clear goals emerged. A couple examples of goals are to “safely enable research and efficient operations through the effective use of people, process and technology” and to “continuously improve computing capabilities to support laboratory operations.” As we fleshed things out further, the theme teams identified clear end states and initiatives needed to achieve them. Some of these initiatives are to “attract and retain talent;” to “develop a financial model for computing in order to understand costs for delivering services, help in quantifying improvements, and to benchmark or compare with other organizations;" and to provide “advanced networking to enable TeraBit networks for exascale computing.” The discussions from these three half-day sessions have been captured in a FermiPoint list which you will soon be able to view.
Armed with all of this information, we are now in the process of assembling the strategy by aggregating a grand vision based on these initiatives and goals. Given that this report is not just for us, it’s important that it provides not just our plan forward but context―how we map into Fermilab’s strategic scientific goals and challenges, what are the industry trends, how computing is an enabler of science, how we deal with regulatory cybersecurity mandates, how we can be more efficient with our investment given our limited budgetary resources and how we expect technology to evolve.
I am excited about this effort. Things are coming together nicely, and we are pushing to have a draft available in the next month. So stay tuned!!!
Ready to race: mdtmFTP released
In 2013, the CCD network research group, comprised of Liang Zhang, Wenji Wu and Phil Demar, was faced with a challenge: they needed to create middleware that would be the basis for a new, faster data-transfer application. Like a cyclist with flattening tires on a smooth road, a data-transfer application will not be as efficient without middleware (full tires). The group therefore came up with Multicore-aware Data Transfer Middleware (MDTM), which would allow a data-transfer application (the cyclist) to take advantage of the multicore system (the smooth road). The team was awarded a three-year DOE grant to complete this research, and since then, they have met with many successes including the release of MDTM, complete with a new data-transfer application and monitoring dashboards.
MDTM schedules and assigns computing resources efficiently based on the needs of the corresponding data-transfer application, which uploads and downloads files to a remote server. The current dominant data-transfer tools in the DOE complex, GridFTP and BBCP, have been successfully used for many years, but new performance and speed expectations, combined with improvements in technology, have pushed open the door for a faster tool. Enter mdtmFTP, a data-transfer application based on the MDTM platform.
In a recent test completed by the DOE, mdtmFTP was able to transfer files at a rate many times faster than the traditional data-transfer applications.
The increase in speed is possible because:
- mdtmFTP uses an I/O (input/output)-centric architecture to complete data transfer tasks. It uses dedicated threads to perform the network and disk I/O operations.
- Various optimization mechanisms, including zero-copy, batch processing, asynchronous I/O, and pipelining, are applied to improve performance.
- The middleware helps maximize parallelism and I/O locality, which means that the data-transfer threads are assigned to cores near the I/O devices they
Lauri Loebel Carpenter
In addition to the release of mdtmFTP, Lauri Loebel Carpenter of the Business Infrastructure Applications group, created mdtmGUI, a dashboard that not only made the project look great, but also allows the Network Research group to monitor how a MDTM-based system is being used.
The network research group is not going to rest on the laurels of the mdtmFTP release for long. “We are also still improving the performance of mdtmFTP,” said Zhang. “We hope in the year ahead we can have it deployed, and our ultimate goal is for other research groups to use our tool in their work.”
Facility Operations/Facilities Support Services
I started working at Fermilab in January of 1984 in the Instrument Support Department of the former Research Division. I came directly from a technical school in Wisconsin, and I started working on NIM, CAMAC, FASTBUS, VME and stand-alone equipment to support fixed-target and collider experiments.
Today, as a data center specialist in the Facility Operations department, my work primarily focuses on the mechanical, electrical, cooling and security functions of the buildings managed by Computing. As task manager, I coordinate and manage laboratory staff and subcontractors that perform work in our facilities to ensure services are in place to provide a strong foundation so Computing can successfully carry out its mission. Such work includes: maintenance and repair of electrical and HVAC systems; security and maintenance of the building and computing rooms; and repair of various types of building infrastructure, all with a continual eye on safety. I am also the Computing representative on Fermilab’s Fleet Utilization Committee.
I am currently taking courses to obtain my Associate of Applied Science degree in facilities management at College of DuPage.
Scientific Computing Services/Scientific Data Processing Solutions/Workflow Management and Provisioning
I am a member of the Workflow Management and Provisioning development group for the CMS experiment. I was one of the initial creators of the application we rolled out five years ago. Since then, I helped with adding more features and supporting operators of the application. Currently, I am responsible for release and code management.
The applications (WMAgent, ReqMgr, Workqueue, WMStats) that our team developed are used by data operators at CMS for day-to-day operations to generate different types of data collected from the CMS detector as well as Monte Carlo simulations. The main functions of the software include automated job creation and submission to batch systems, such as HTCondor, by given configurations. The software also helps track the data and monitor the progress of the jobs as well as support a retry mechanism. The core library is also used by other CMS applications such as DBS (Dataset Bookkeeping Service), SiteDB, (user and site db), CRAB (a user analysis data submission and tracking system).
In terms of future projects, I am involved in developing software performance analytics of archived data such as job performance and resource usage statistics.
I joined Fermilab in 2007. Most of my prior work dealt with monitor application development for the NOvA project.
Lesson learned: Employee shocked by damaged cord
While going about our daily tasks, we sometimes forget to take the proper precautionary steps to ensure our safety. One example of this happened on June 24 when a Los Alamos National Laboratory employee received a shock while attempting to raise a Baker electric ergonomic workstation.
The employee reached below the ergonomic workstation to check if the cord was plugged into the outlet box located below the workstation. The employee did not have a clear view of the outlet box to see if the cord was damaged or securely attached to outlet box and received a shock while attempting to check that the cord was plugged into the outlet box. The employee immediately reported the event to the supervisor, who contacted a deployed Industrial Hygiene and Safety professional. The workstation was taken out of service and all workstations in the group and directorate were immediately inspected.
How to reduce the risk of electrical shock:
- Before plugging or unplugging electrical equipment, perform a visual check of the cords and outlets for damage.
- Employees and responsible line managers should make sure they have a clear view of electrical cords and outlets in their areas and should inspect their work areas to ensure cords and outlets are not damaged.
- Workstations with damaged cords or damaged outlet boxes should be taken out of service and replaced.
Now playing in the FCC lobby:
"OCIO all-hands meeting," Rob Roser, OCIO all-hands meeting, Aug. 18
"Exascales and Exabytes: Future directions in HEP Software and Computing," Oliver Gutsche, meeting of the Division of Particles and Fields of the American Physical Society, Aug. 6
"Framework Introduction," Marc Paterno, art/LArSoft course, Fermilab, Illinois, Aug. 3.
"Scientific Workflows Using Science Gateway Technology," Saba Sehrish, XSEDE Gateways and Workflows Symposium Series, March 13.