Abstract

CSDE currently operates a student access computer lab in room 106A of Savery Hall. This 173 square foot room contains 8 Dell Precision workstations and a large capacity network printer, all of which were purchased under previous STF grants. CSDE wishes to expand the resources made available to students by moving into a larger space (Savery 106F), purchasing five additional workstations, upgrading the current machines and installing the latest software on each.

Background

The Center for Studies in Demography and Ecology is interdisciplinary, supporting UW students and faculty interests in demography and ecology. Demographic research covers a broad spectrum of topics, including issues related to health, AIDS, crime, welfare policy, family, sex roles, race, ethnicity, segregation, gender, inequality, poverty, family planning, and environmental issues. At present, the Center has affiliate students from many departments on campus including Anthropology, Economics, Geography, Political Science, Sociology, Public Affairs, Social Work, the Jackson School, and the School of Public Health and Community Medicine. Most demographic research is very computer intensive, and the field is one of the most quantitative of the social sciences. Success in the field requires demonstrable computer literacy, and a record of publishing quantitative work. Many data sets used in demographic research are very large, sometimes involving millions of cases. For example, the 1990 1% Public Use Micro Data sample (1990 Census) has 1,105,583 household records and 2,500,052 person records. Students may also use the a 2% sample of the Chinese Census, which is far larger. Processing large data sets without high speed processors and very fast I/O is very time consuming. Increasing I/O and CPU speed makes a huge difference in the amount of time it takes to process large data sets.

We are privileged to serve a variety of non-traditional students, and have received funding from a variety of sources to allow us to do so. The David and Lucille Packard Foundation and the Bill and Melinda Gates Foundation have provided funding for the "Population Leadership Program". (See http://csde.washington.edu/plp.html) This program solicits applications from professionals from developing countries working in population policy, family planning and reproductive health. The program aspires to provide persons with the skills, knowledge and support that can sustain them as they rise to higher levels of leadership in their countries. Fellows in this program are provided with a stipend, tuition, and the services of the Center.

American minorities are under-represented in the field of demography. We received funding from the Melon Foundation to encourage minority undergraduate students to develop interest in demographic research. The funding for this program recently ran out, but, for several years, every summer minority students would be recruited for this special program. They would work closely with a faculty member affiliated with the Center, and would have full access to the resources available in the Center. Some students who participated in this program are still active in the Center, and use our computing infrastructure. It is a priority to the Center to attract and train undergraduates in the field of demography.

It is part of the Center's mission to help University of Washington's demography students attain success in their field. To that end, we provide a varied range of support, including:

1) A weekly colloquium series. (See http://csde.washington.edu/ and click on Seminars.) We use this forum to bring in respected national and international demographic scholars. We also use it to provide opportunities for students to present their own work. This series provides an extremely valuable service to students. It allows them to see first-hand the work of scholars in their field, and exposes them to the breadth of the discipline. At the end of each colloquium, time is set aside for students to meet with the scholar. Meeting other scholars, hearing their work and seeing their presentation style, is an important part of learning the craft of demography. Most students affiliated with the Center hope to continue research and possibly teach in the field. An important part of the "craft" is to be able to present their work well. Exposure to the ideas presented, as well as to the presentation styles, is extremely valuable to students. Presenting work of their own will be an important part of their own career.

2) We provide a library and library support to assist students with their research. However, it is a small library, and cannot provide all of the resources needed for research.

3) We provide computing infrastructure (partially provided by previous STF support).

4) We provide consulting in computer use, computer programming, statistical methods and help with specific software packages. Among others, these software packages include SAS, STATA, SPSS, and SPLUS. Learning programming software commonly used quantitative social science is very important to the future careers of students in demography. Computing in demography is often data and CPU intensive. Speed of I/O and CPU makes a big difference in the productivity of students.

5) The Center also provides informal support that is the beneficial by-product of people working together. The Center is fortunate to attract students with diverse backgrounds and diverse interests. We serve students from many departments, and from many countries. This diversity provides a synergy of ideas and cooperation that serves the students well. Students consult with each other as they work on projects, and help each other as they work on class projects, theses, dissertations, and prepare for professional meetings and job talks. The cross-fertilization between cultures and disciplines is extremely valuable. Having a lab with up-to-date hardware and software is critical to this mission.

We currently have 106 students on our mailing list, and the number is growing. Any student with interests in demography can become an affiliate and use Center resources.

Benefits

CSDE's facilities, including the center's computer lab are devoted to demographic research, and the needs of students we serve. Here students can work on projects for their classes, theses and dissertations and get help from Center staff and faculty located nearby. This combination distinguishes CSDE's computer lab facilities from other student access computer labs on campus.

Also, many CSDE students teach classes of their own, and use our lab to prepare their classes. Teaching will be an important part of the careers of many of our students. As they prepare classes, working in a lab where there are other students with similar interests and experience is very helpful.

General purpose computer labs do not have powerful workstations, sufficient storage space, the appropriate software or the consulting ability for demographic work. In CSDE's lab, students have access to machines and can store their data, access their email, and work on papers in one location. CSDE also provides services that distinguish it from CSSCR. First, CSSCR serves general student computing needs in the social sciences whereas CSDE's lab focuses on the large data requirements for demographic analysis. Second, unlike CSSCR, CSDE provides 24 hour access to its computing facilities for those students affiliated with the center. And third, CSDE provides specialized software and consulting on demographic issues and students have quick access to the CSDE library across the hall.

The ability to do cutting-edge demographic work depends on access and manipulation of large datasets. In our year 2000 STF proposal we addressed the problem of insufficient network storage space for student files. With this addition the lab's functionality has been greatly increased. Consequently demand for access to lab resources has grown.

Purchasing additional workstations and the requisite software and upgrading the machines we currently have will allow us to serve more students and more fully utilize the computer resources provided in previous years. The current workstations need memory and drive upgrades so they can continue to remain functional as the demands of software become greater. The current configuration of 64 MB of ram and a single 4GB hard drive is inadequate for the Windows 2000 operating system and the statistical software we wish to provide. The server allows safe, backed up storage of large data sets, and, it allows data to be available to any machine on the network. But processing large data sets from the server is slower than processing from the hard disk on the computer. With larger hard disks on the computer, students will be able to copy their large files from the server to compute locally. This will increase their productivity, and improve their work environment.

To estimate the improvement in performance that the upgrades and new equipment will make, we ran some tests. We used SAS, a program used by students at CSDE, to read raw census data for Washington State (Summary Tape File 3A data, ICPSR Data set 9782) select only King County tract level data, and save a SAS systems file. This is not a very large data set compared to many used by demographers. Only 68,604 records were read, and 529 observations were saved. It is not uncommon for demographers to process millions of cases. Slow processing and I/O is a burden that greatly reduces the productivity and, thus, the future job opportunities for demography students. The time savings of running the SAME job on a faster machine, and with local I/O instead of accessing data from the network server is huge. On a newer machine with a 933 MHZ processor and 512 MB of RAM that is on the desk of a CSDE staff member, the job took 49 minutes of real time when reading and writing to the network server, but only 1 minute and 4 seconds when reading and writing to a local hard disk. On one of the machines in the lab, which has a 400 MhZ CPU, 64 MB of RAM, the job took over 46 Minutes when the data were read from and written to the server, and 2 minute and 28 seconds when the data were read from and written to the local drive. The local drive was almost full, and would not have been able to store a larger data set. This job is intensive with I/O, but is not intensive for CPU. We also ran a program that required less I/O, and more CPU. This job read four variables from the data set created in the previous job, and estimated one multiple regression equation with three independent variables. Multiple regression is a statistical method frequently used by demographers. The more independent variables used, the more CPU time is required. It is not uncommon for demographers to estimate more than 50 regression models a day, with ten or more independent variables as they try to understand what the data has to tell them. There are many statistical techniques frequently used by demographers that are far more CPU intensive, but, since this is a very common technique, it is used for illustrative purposes. The real time for this program ranged from 17.01 seconds to 3.82 seconds. Someone might say that waiting even 17 seconds for a program like this to run is acceptable, but, remember that students in CSDE may run hundreds or thousands of these while they are in their program of study.

I/O Intensive Job using data from network
Lab Machine: 46 minutes (real time), 3 seconds (CPU)
Staff machine: 49 minutes (real time), 1 second (CPU)

I/O Intensive Job using data from local drive
Lab Machine: 2.5 minutes (real time), 1.75 seconds (CPU)
Staff machine: 1 minute (real time), 1 second (CPU)

CPU Intensive job using data from network
Lab Machine: 17.01 seconds (real time), 1.14 seconds (CPU)
Staff machine: 8.68 seconds (real time), .24 seconds (CPU)

CPU Intensive job using data from local drive
Lab Machine: 9.92 seconds (real time), 1.17 seconds (CPU)
Staff machine: 3.82 seconds (real time), .28 seconds (CPU)

Network access would vary depending on network traffic. These tests were run on 12/26/2000 when network traffic was extremely low. In normal daily activity we would expect a higher load on the network with a consequent impact on through put.

Student Access

Any student on campus with an interest in demography will be given access to the server upon request. At present, students who want to use CSDE resources can get an account on the center's network and begin to use the computer lab. The lab is open during business hours and students who affiliate with the Center can get keys for 24 hour access.

Available Resources

SERVERS

Current
Dell Poweredge 2100/200 server used for faculty, student and staff file storage as well as print spooling and network administration.

Dell Poweredge 4300/400 server used for faculty file storage

Dell Poweredge 4400/800 server used for student file storage (previous STF grant)

Micron PC used as a departmental web server

A Dell Powervault DLT autoloader for tape backup of all servers

Planned
A Dell 705 NAS to be used for faculty file and storage. (Ordered, delivery expected in February)

Two DEL Poweredge 1550 servers to be used for terminal services and permit remote processing of large data sets. These will be available to students and faculty. (March)

UNIX server, yet to be configured. (Will be available to students.)


STUDENT LAB (Partially funded by STF fees, and partially from other CSDE funds)

Eight Dell Precision 410 workstations, 64 MB of ram, 400 Mhz processor, 4GB hard drive
HP scanner
HP 8000 network printer for student use

OTHER EQUIPMENT (Also sometimes used by and for students)
Current
HP 4000 printer for the library
HP 4050 for a CSDE funded research project (THEPS lab)
HP IV for the front office
Scanner for the library
Various computers in staff, faculty and student offices
Computers for CSDE funded research project (THEPS lab)

Planned
We are planning on purchasing a network color laser printer.

The new workstations will be network ready, and can be used to expand our existing on-campus infrastructure. Most afternoons, every work-station in our lab is in use. Additional machines will allow more student to access CSDE computer resources

CSDE will share the cost of the project by paying for additional tables, chairs and network ports, and insurance on the equipment.

Installation Timeline

As soon as we get the funds, we will order the equipment. As soon as the equipment comes in, it will be set up and made available for use.

Departmental Endorsement

CSDE is a multi-disciplinary Center with affiliate faculty and students from Anthropology, Economics, Geography, Political Science, Sociology, Public Affairs, Public Health and Social Work.

Faculty have written to provide information about how the proposed equipment would help students affiliated with CSDE.

"The CSDE computer lab is vitally important for the professional training of CSDE graduate students. It is the setting in which most students learn to use the statistical software packages that they will use throughout their professional careers. It is the setting in which they will compose their theses and first professional papers. It is absolutely essential that the CSDE computer lab offer students a first-rate environment within which they can be professionally socialized in these critical activities. Technology changes so quickly that many of the computers in the CSDE lab are out-dated. Student demand has increased, and it is more and more common to see all of the computers in use. Upgrading the lab equipment and increasing the capacity of the lab would be very important steps toward assuring that our students have access to the best computing infrastructure for their professional training. Other universities are providing their students with the best computing environments. If we don't follow suit, then our graduates will be at a disadvantage when they compete for jobs."

Upgrades to ethernet and electrical systems that will be necessary will be paid for by CSDE (possible with some help from the Dean's office). CSDE will share the cost of the project by paying for additional tables, chairs and network ports, and insurance on the equipment.

Student Endorsement

Our current lab is insufficient to meet the needs of the students we serve. We asked students what would help them get their work done in the CSDE lab, and, specifically about our current lab. Here are some of the comments we got.

"Within the past year, the software has been upgraded in the CSDE computing lab. New statistical packages were added to some machines, and older software versions were upgraded. These are great improvements, but they draw attention to our computer hardware deficits. These machines simply do not have the processing speed to take full advantage of our improved programming capacity. As one who works with large data sets, I am frequently frustrated by the time required to process fairly simple analyses. Waiting a few minutes for each analysis to be completed quickly adds up to hours of time wasted.

As one who uses the CSDE lab on a daily basis, I have noticed that the lab has become increasingly popular over the past year. Students arriving in the afternoon often find there are no machines available. The fact that some software packages are only available on a few machines further compounds the problem.

The space and heat problem in the CSDE lab is quite serious. I find it difficult to use the lab for any type of collaborative work with another student, because there is insufficient room for two people to work together at one work station. Because we are squeezed together so tightly, it also feels disruptive to have any type of conversation. The lab really is not conducive to collaborative work and yet working as a team member is an important professional skill to develop."

Another student wrote:

"I'm writing this as I wait what seems like an eternity to run frequencies on one variable in a 2,000,000+ case dataset. Besides slowness, I consistently run into problems with lack of processing/storage space whenever I use a large dataset.

More speed and space are definitely my top priorities. Two more computers (so 10 instead of 8) would be nice, but it's already really hot in here and this would just make it worse.

Re. software, what you propose looks good. SAS on all machines? What about ArcView?"

And finally - a few of us were wondering whether a technology grant could cover some new chairs and staplers."

{Author's note: This student mentioned only two machines because she did not know that we have the option of moving the lab to a larger room if we get this funding. And, we have put staplers in the room, and will upgrade the furniture.}

Another student wrote:
"Hardware issue:

I find that the machines in the CSDE lab are very slow to start and/or log on to the system and to run programs. These are slower compared to my own home computer and the computers in the sociology lab. They are faster than the CSSR lab computers though.

Software issues:

It would be very handy for us if all the computers had all the statistics softwares installed in them. For instance, some selected computers have splus installed while the others do not, which means one has to get hold of those computers to do the work using that program. Perhaps it contradicts with your indication of stat package preference.

Currently I am using SPSS and SPLUS. In near future I am hoping to use STATA as well."

Another student wrote:
"I've only been using the CSDE lab for a brief (!) time :), but here is my initial impression about hardware:

If new pc's are purchased, it would be helpful to have zip-drives on each new machine. Also, the computer mice with rollers on top help me move around large datasets without putting undue stress on wrist, etc.

The software that I use is as follows:
Stat packages:
Stata
Splus *including spatial modules*"

Another student wrote:
"One of the biggest problems with the CSDE student computer lab is the room temperature. The room is poorly ventilated and stifling hot. It is particularly difficult to work in the lab during the summer. In the summer of 1997, I ... spent every afternoon working in the CSDE computer lab. I was also in my third trimester of pregnancy. The temperature in the room was extremely warm, and I was required to leave every half hour to cool down."

"Yes, it does get uncomfortably hot in the lab sometimes. It would be great if we could get cooler equipment."

"Thanks for your concern about condition in the lab. Even in winter, it's so warm that shorts and a t-shirt would be the most practical attire. In summer it's really a sweatshop. Flat screens that give off less heat would be a wonderful improvement."

Another student wrote:

I think that the proposed plan for utilization of STF fees sounds great, I would add that a place for students to create data cd's would be useful (otherwise we will be keeping you very busy :), access to laptops for presentations is an excellent idea. In addition to adding and upgrading equipment, the plans to make the lab a bit more comfortable are much needed--the heat and the cramped space of the lab can really be a deterrent to getting work done. I think many of us like the collective atmosphere of the lab, and not working in isolation, but sometimes it is just too crowded for comfort in the lab.

Items

Below are the items making up the current proposal. The asterisk (*) beside items signify that they were approved by the committee. This however was not implemented correctly for our database before 2005, so earlier years may not show this.

Click an item's title to view details on that item, or show all item details.