Making Network Maps
Jesse Davis' Map Experience
Sunny Christadoro - Map Wars
Brandon Oelling Fights On
Joshua Harley Continues the Battle
Chad Shaffer - Dynamic Changes
Jesse Janowiak - Another Jesse Takes Over
Eric Johney and the Great Network Map Adventure
Chris Hurlburt
Jesse Davis' Map Experience
1976-1988
Back in August I agreed to pictorially map the history of the Engineering Computer Network. I had no idea what I was really getting into. Sketches and printed information were provided to help get the project started. It looked like a "piece of cake".
The technical drawing of the symbols was not difficult. The real trouble was trying to create symbols from verbal descriptions without knowing what the actual computer looked like. "What does this or that look like?" was a frequent question of mine. Much time and brain picking was required to get the 1976-1986 maps completed. "Which machine was hooked where?", "What MIPS designation should be on that machine?", or "How many terminals were on that network then?" were but a few of the numerous "layout" questions I needed answered. After many revisions and check plots, the first ten years of the network growth were chronicled.
However the hardest part was still to come, putting together a map of the current system. It was soon apparent that a graphic representation of what "was" is easier to capture than an illustration of what "is". As a beginning point for today's ECN, I was given an April of 1988 map. I knew I was in trouble when I began asking questions based on the April map and got the replies like "Where did you get that old thing?" after the ECN guys finished laughing. This was only August and they were calling an April layout "OLD"?
I soon came to realize that the April map was "OLD" because the new engineering building opened this summer followed by the "great computer shuffle" between EE and MSEE. With so many changes I decided to trash the "OLD" April map and get the people "in the know" to give me an OVERHAULED version of what is where. Finally it was December and the 1988 poster was mostly finished. I went home for one weekend and came back only to discover that there had been more computers added to the system. This is the dynamic state of the entire network. Common replies to my "layout" questions were "On what day?" or "What time of day?". This project made me realize just how fast things are changing on the Purdue campus.
Sunny Christadoro - Map Wars
1991
In May, I was asked to complete the "final" alterations to the 1990 map, which was under modification by Steve Weinrich, an intern at the ECN. It was believed that most of the changes had been completed and only a few minor revisions were necessary before the final draft could be produced. Unfortunately, this was an incorrect perception. In reality, with a map this complex, minor changes require major reworks.
So, I began the imposing task. Upon revising the update on the AutoCAD Release 10 program, I returned to the ECN office. Immediately, the ECN personnel appeared with red pens in hand and attacked the map by adding, deleting, and correcting any discrepancy. The result was a chaotic mass of red lines. To my dismay, this scenario occurred often!
Back to the drawing board, or should I say terminal screen, I went. Here I would try to decipher the inky scribbles. Weekly I would post "latest & greatest" map and hope this was the last revision, but the corrections and the confusion never seemed to end. Finally, the day arrived when everyone agreed that the map was accurate and complete. I breathed a sigh of relief, thinking that the difficult part was over and that I would soon be finished with this project.
WRONG!! That brief euphoric feeling evaporated as I tried plotting the map on the Hewlett Packard Draftmaster in the Agricultural Engineering building. I was plagued with problems. The pens and paper would not cooperate with one another At first, the plotting pens' ink was too faint. which produced a bleeding ink, blurry text, and in some unique areas "invisible ink." After much time and frustration, this obstacle was overcome.
So five weeks later, the project was completed and it was time for me to depart Purdue and go home for a quiet summer vacation. Rest was necessary in order to recuperate from my higher blood pressure, tension headaches, and hair loss (I had pulled out most of it while working on the project.), all caused by these so- called "final" alterations.
Brandon Oelling Fights On
1995
In 1976 the ECN was a single machine; today it is over 2,000! The most recent map is near completion and will soon be ready for display.
After many months of research, drawing, and revisions, the new 1994 map is totally revamped. Since production of 1992's map, the ECN hardware staff has not only added new equipment, but redesigned many departmental network configurations to take advantage of faster and more efficient machines. With faster machines having been added to the network, most notably Sun SPARCstation 1000s, a new MIPS (Millions of Instructions per Second) color has been added to the key.
One important lesson I have learned since taking on this project is that the network never stops growing and it never gets a rest. Comparing the interim 1993's map with 1994's map, it is immediately apparent that the network has grown extremely fast in just one year. A department that used to need two servers and a few workstations now requires four to six servers and dozens of workstations to meet computing needs. With programs like Mosaic and Netscape making Internet browsing easier, more and more people are using the network and as expected the use of and demand for network access continues to rise.
Joshua Harley Continues the Battle
1997
At the end of 1996, the network consisted of over 2,700 machines!
Since 1994, the network has continued to grow and undergo some major changes. The network backbone has been upgrades from an Ethernet with 10 Mb/s to an ATM with 155 Mb/s transfer rate. Sun Microsystems' Ultra 1 and 2 technology has also been incorporated into the network.
During the spring semester, I spent a lot of time researching the network to find all the machines attached to the Engineering Computer Network, (i.e. all machines with IP addresses starting with 128.46.XXX.XXX). The hardware staff lead me to a key network file that lists all of the computers connected to Purdue's network. After spending many hours going through this file and making many revisions to the AutoCad drawing, I believe that this is one of the most accurate maps to date.
Chad Shaffer - Dynamic Changes
1998-2000
The one constant throughout developing both the 1998 and 2000 network topologies is that as computer technology changes so then will Purdue's Engineering Computer Network. For that reason, the network maps are an instantaneous "snapshot" of the ongoing improvement of computing facilities available to engineering students and faculty.
Jesse Janowiak - Another Jesse Takes Over
2002-2004
2002
Chad graduated only a few weeks after finishing the year 2000 network map, leaving a very short time to train me in his techniques and a very long time before I could start working on the 2002 map in earnest. This was unfortunate because I was not able to benefit very much from his experience with the map. However, it was also a blessing in disguise. Because I was not trained in detail how to complete the map, I was forced to figure out the best way to go about it from a fresh perspective. As I examined the steps that needed to be taken, it occurred to me that a great deal of the process which had previously been done by hand could be automated with a little help. When the time finally came to begin work on the map, I had the assistance of Mark Senn, one of ECN's system programmers. He wrote a program that searched through ECN's host file and gave me totals for every machine type on every subnet. Josh Harley and Chad Shaffer after him had to count these tallies by hand, and Mr. Senn's program saved me days of work.
When I had neared the end of the project, one of the last steps was to get new text for the bottom of the map. The text is supposed to explain the most significant changes to ECN since the last network map was printed. For this I turned to William Simmons, director of ECN, who gave me a few sentences that summed up the changes. As I was adding the text to the map, something struck me as odd and I returned to his office. "Mr. Simmons," I said, "Your description says that in 2001, ECN's total number of computers reached 3,500, but my count is closer to 5,000. In fact, even the year 2000 network map displays more than 3,500 machines." He agreed that the discrepancy was a problem, and he showed me the source of his statistics. I was surprised (and a bit dismayed) to learn that he was implementing a new machine counting method that is much more accurate than mine since it automatically searches the network for active computers in real time instead of relying on a static database that is maintained by hand. Mr. Simmons and I put our heads together and realized that the difference of about 1,000 machines between our counts was the result of machines remaining in the database even though they are no longer actually present. Mere days before I was to print the final copies of the map, I had to redo the count from scratch using the new data provided by Mr. Simmons. Fortunately, all of the subnets were in the right place already, so it was a matter of several hours' tallying and correcting before I finished.
The most interesting change in the map visually speaking is that for the first time in the history of ECN, this map is smaller than the previous map. The reason is not that ECN has fewer computers (although the improved machine gives that illusion). The smaller map is due to simplification and consolidation of ECN subnets. Where there was once three or four subnets, there might now be only one because the subnets were merged. What started out as a messy scattering of servers and workstations in the early 1990s is becoming a centralized and organized unit.
Of the few contributions I made to the network map, my personal favorite is the changing of the Mac icon. The Macs in ECN have been represented as 1992 model machines for way too long, and it was time for a change. I am very pleased with the result.
2004
As much to my own surprise as anyone else's, I was around for the 2004 network map as well. The 2004 network mapping process was especially complicated. When I began working on the layout for the map, I interviewed site specialists and ECN technicians to get an idea for where the connections needed to be drawn. The most common comment I received was "here's how it is now, but that's all going to change soon." The BoilerNet project, a campus-wide initiative to improve connectivity at Purdue, was currently underway, and it was making my job very difficult. ECN was in a state of transition in preparation for the new campus computing architecture, and I couldn't get a solid answer from anyone about how their subnets were physically connected to the rest of campus.
Finally, someone advised me to speak with Larry Billado, who was involved with the BoilerNet project. He was able to answer most of my questions about how things were going to be laid out under the new architecture. Armed with that information, I decided to proceed not with a "snapshot" of the network as we've done in the past, but with a projected estimate of where the network would be by the time the map was finished.
As I write this account, copies of the 2004 network map are printing behind me. Although I tried to be accurate, the map remains merely estimates of the current state of the network. I left my own touches on this map, primarily in the form of all-new icons for workstations and an improved, grid-based layout. I hope this map serves as good reference to future network historians, rough though it may be.
Eric Johney and the Great Network Map Adventure
2006
Thanks to Jesse's great work on the 2004 map, I was able to use it as starting point and build on top of it, since no major changes in the network structure had taken place. The majority of changes that I made to the map involved reorganizing everything into more of a grid layout to make room for new subnets and servers that had to be added.
One of the problems I ran into was that the map needed to be grouped by school; however, the layout has subnets connecting to various building routers. The problem comes in when one school has computers in a completely different building hooked up to a different router. Rather than having lines crossing each other a bunch of times all over the map, or getting rid of the current layout, I decided to simply denote that the subnet was located in a different building in the subnet title. As I understood it this was a fairly trivial detail, since they all connect to the rest of the Purdue network anyway.
Overall, my mapmaking experience was a good one, as I am happy about how it turned out. The Engineering Network continues to grow, and the mapmaking continues to simplify. Finding the information about what is actually connected to the network is still not easy, but I believe it has turned out to be fairly accurate.
Chris Hurlburt
2008-2010
Chris Hurlburt took over responsibility for the ECN Network maps for the period of 2008 - 2010. Due to the virtualized nature of networks the one to one correlation between lines and wires had broken down. As a result a rethinking of the network map structure took place. Computers were organized by unit rather than physical router. As a result some routers appear on the map in multiple locations.
Chris created updated icons for nearly all of the computer types shown on the map. Line structures were also changed due to the density of computers.