Bootable Cluster CD:About

From Bootable Cluster CD

Comments on the BCCD Project from the Project Leader, Paul Gray

What started off as a little curiosity has certainly snowballed into something well beyond what I had originally envisioned. Now it seems as if there is no end to all of the niches that the BCCD can address. This is not a blog -- sorry...just a personal stream-of-thoughts that I've been wanting to convey to anyone that's interested enough to explore this page.

Motivation

One thing that has not changed throughout the ups and downs of the project's development has been the underlying motivations which drive the project. The original motivation is still the strongest, namely the BCCD project's main goal is to promote High Performance Computing Education (HPCE).

However, many (myself included) have used the BCCD image for other purposes (distributed password cracking, a system recovery CD, as a drop-in web server, as a wireless network monitor, dropping in clusters at Cicuit City, etc.).

Looking back at the origins of the project, there once was a time when clustering resources were very hard to come by, extremely costly, difficult to configure, and impossible to maintain. (Note that some might also argue that this is still the case.) We therefore run into the chicken-and-the-egg problem of how does one train undergraduates in HPC without access to HPC resources?

History

Late in the year 2001, I stumbled on the "LinuxCare" bootable CD. The "LinuxCare" BBC was a wallet-sized bootable recovery CD put out by the folks at "LinuxCare." The "LinuxCare" project forked into projects with different intentions, one of which I latched onto (lnx-bbc) as a contributing developer, where to this day I contribute to the development of the ultra-small, bootable recovery image. Involvment in the LNX-BBC project has been very symbiotic. What I've learned from them has been immeasurable.

Sorry. Back on topic. Stealing an often-posted comment from Slashdot, back in the "LinuxCare" BBC days I was left thinking "Imagine a cluster of these." Wow. With a cluster of these, you could walk into a room full of PCs running that "other" operating system, and instantly have yourself a clustering environment. Ever since then, this has been my obejctive: to be able to support classroom instruction and research through the use of open labs of workstations; allowing them to remain unmodified without dual booting, nor needing to modify the hard drive in any way. This is the embodiment of the BCCD today.

Philosophy

One thing that cannot be quantified is the pure enjoyment returned to me through this project. Without question, building and tuning the BCCD has brought about a multitude of problems that needed to be solved, has taken me to numerous obscure works of other individuals, and has overwhelmingly reinforced my enjoyment of distributed computing environments. Hopefully some of this enjoyment has been infectious.

Direction

"Wow." That's about all that I can say. There is always the day-to-day minutia to address with regard to supporting the latest and greatest (and not-so-greatest) hardware. But this type of work doesn't motiveate me on a daily basis. Where is HPC going in the near future? I have all of these ideas about supporting Grid services with the BCCD, using the BCCD for temperal clustering environments, I even have heavily tweaked-out versions of the BCCD image that I use for teaching the System Security course at UNI. A pure 64-bit x86_64 BCCD? (Probably not worth the port, sorry.) The reality of things is that development of the BCCD has been primarilly governed by YOU. What do you need to support HPCE? What software and tools do you need to peg the "gee-whiz" meter (yours, students, colleagues, or even your spouse's). Keep me posted on what you want, and I'll promise to keep motivated, and to keep this project moving forward.

Appreciation

Two individuals need to be credited for extending support to this project. I am certain that the BCCD project would not exist in its current state if not for Bob Panoff at the National Computational Science Institute (NCSI) and for the dedication and work ethic of Sarah Diesburg. I can only strive to attain the enthusiasm that is embodied in Bob Panoff's organization: Bob's support of my project allowed me to focus on producing a production-quality product, which I believe we have done. In addition, I cannot overstate my appreciation to Sarah Diesburg for her dedication, insight, and support extended to this project. I roped her into this project with a security smokescreen, luring her into setting up some of the server's services. While she doesn't share the same zeal for MY project, it's clear that she's been bit by the same bug and will do exceptionally well in graduate school (where ever that may be)!