Sometimes we just fill in gaps as we go along. For instance, Shaon’s talk, “Next generation cloud deployment: Self help is the best help!” just got accepted to SoftExpo 2012. Subsequently, a few of us had a spontaneous discussion on how to field a couple tricky questions when giving EuCa talks – the post below is based on these meeting logs. (Disclaimer: these represent individual opinions and don’t speak for Eucalyptus as a company, etc.)

Why is Eucalyptus using the Amazon API?

Because AWS is way farther ahead than the others, and we believe it to be the most worthy of focus. We may support other APIs in the future, but we prefer to focus now on the best API, and that’s AWS. See this post by Mark Shuttleworth for reference – it’s about OpenStack, but makes the same point.

How much does Eucalyptus support cost?

Varies widely. Contact Eucalyptus for details.

What third party apps are available for Eucalyptus management?

There are a few options for apps that run on Eucalyptus and provide Amazon-RDS-like support. The top three:

• HybridFox
• RightScale
• Enstratus

Providers of IaaS services need a usage monitoring tool. What does Eucalyptus offer?

There are hooks in Euca 3.0 for reporting usage; this is a new feature. We don’t have details yet, but they will be in the docs when we release them (which will hopefully be soon).

In Euca 2.0, instances cannot be recovered once the node goes down. Ks there any solution?

Boot from EBS is a new feature in Euca 3, and may be able to help with that problem somewhat. However, redundancy (multiple instances running at all times) is the way to survive instance failures. If you have multiple zones, start instances on all zones.

Instances are supposed to be spendible. HA (High Availability) is for the infrastructure: the app will still need to be designed with the cloud in mind to be fully HA. EBS and Walrus are the persistent storages to be used to keep the states and backup.

Wait, wouldn’t EBS and Walrus use more resources? I thought cloud was supposed to minimize resource usage.

Yep. HA wastes resources by design.

What are the alternatives to Eucalyptus, and why would someone choose Eucalyptus over them?

• Openstack: Broad community supported by many companies, modular design. Both a blessing and a curse. Openstack is much more a set of tools for building a cloud; Euca is cloud-in-a-box. And openstack, honestly, just isn’t as far along.
• Cloudstack: Good product, good UI. Integration with AWS isn’t as good.
• Opennebula: Again, more of a toolkit approach. Not too many components — just more flexibility about how you put them together. Seems like Openstack and Opennebula are both good for the service provider market, and Euca and cloudstack are more all-inclusive products for the enterprise market.

That’s all we had time for — what other questions would you ask, and how would you answer these?

I have decided that the POSSE owl should be named Superb Owl. So that he can have one day a year dedicated to him.

(Note that the POSSE owl is not necessarily a male owl, and that my posting of Robyn’s quote does not constitute my endorsement of such a name for said owl. Still, it made me stammer incoherently for a moment. Thank you, Robyn.)

That is all. Have a very good evening, and may your nacho overdoses be miraculously healed with a good night’s sleep.

A marshmallow was offered to each child. If the child could resist eating the marshmallow, he was promised two instead of one. The scientists analyzed how long each child resisted the temptation of eating the marshmallow, and whether or not doing so had an effect on their future success. (Source: Wikipedia)

Short answer: yes.

Greg and I were talking about the difference between production and production capacity a few days ago, and the importance of balancing the two. It’s not a hard concept; we do this all the time when we play video games. When you play Monopoly, you build houses and hotels because you know that’s going to give you the strong resource and financial base you need to wipe the board with everyone else at the end; when you sit down for Settlers of Catan, you build cities – you don’t just start hurtling roads out there, right? You want that grain, that ore, those bricks. You want that power at your fingertips, so you Do The Marshmallow – you focus on building that power, even if it means not using all the little bits of power you have right then. Less shiny now, more shiny later.

What does this have to do with FOSS? Well, I’m reminded of the Marshmallow Experiment every time I see something like this:

“Linux geeks not caring about noobs is the main reason Windows is so popular.” –Chris Watkins

That’s my friend Chris, from Appropedia. Chris is a technical guy who loves the Free world; he’s an engineer working on disseminating open-licensed appropriate technology information to grassroots communities of hackers in the developing world using an entirely open-source software stack to do so. His statement reads to me like a bug report on FOSS’s ability to build production capacity in its communities. (We’ve gotten better, thanks to tons of long, hard work by many different groups and people – but there’s still a long way to go.)

I am also reminded of the Marshmallow Experiment every time I see something like this:

 I just think it’s bizarre. “We need more people! Lets try to recruit those with this particular type of sex organs!” –from a GNOME Women comment thread

Dude. Do you want to curse the darkness? Or do you want to light some candles? Because what you did right there is called “cursing the people who are lighting candles.” When you see someone trying to improve the capacity of a community you care about, try helping them. Constructive criticism is helpful; however, the above comment is a good example of destructive criticism. Here’s how to tell the difference.

These comments are, in different ways, both about building production capacity in FOSS communities. In a world where software is considered obsolete after a year or two, where 6-month releases are built in no small part upon the outputs of 48-hour hackfests, where there are so many compelling reasons to focus on the now – what does your project do to look into the future? (Does it?) Could you see those two scenarios above applying to the communities you work within?

We’re qualitative researchers, and we do a lot of interviews – for instance, talking with faculty who teach open source (me) or engineers on interdisciplinary teams (Robin and Tiago) or people who mentor minority STEM students (Joi). Whenever we do an interview with someone, we always send them their transcripts and our writeups on them afterwards to catch mistakes and make sure they’re okay with what we’re saying about them in our papers. Privacy and confidentiality are vital.

And that’s the problem. Security is in ZOMG mode, so we can’t simply stick these up on Google Docs, which was one suggestion – creating individual password-protected posts on our WordPress instance was another, which I instantly flinched at and nixed. For similar reasons, Dropbox was a no – not to mention I had nightmares of how that would (not) scale. Having one person email every subject is a nightmare, even if we use a separate group email account; inboxes are not made for workflow management, let alone collaborative workflow management.

And so I turn to you, o internet, for any ideas you might have. It’s safe to assume everyone involved has email and internet access. It’s also safe to say they’re probably not computer geeks and that any solution needs to be web-based or cross-platform. We can get hosting space and stand things up ourselves, but don’t have a dedicated professional sysadmin (I’m the closest thing we’ve got) and don’t have much budget – if we have any at all – to throw at services. I may be able to beg IT to do something for us if they already have instances of that software running for other people, but there are no guarantees.

What would you do? Here’s what I’ve thought of.

1. Drupal. With customization, you can have any content workflow ever known to mankind. That’s a ton of customization, though – and we’d need to make unique logins for everyone, which is a pain (we don’t want to force our interviewees to acquire Yet Another Web Login if we don’t have to). IT will know what Drupal is, but I’m not sure they’ll want to babysit something so heavily customized.
2. Etherpad. Opens up the option of collaborative transcript editing and easy viewing of a writeup’s edit history, which might be handy especially for remote (phone, but sometimes videoconferencing or text chat) interviewees. Security is sketchtacular, though; you can protect each pad with an unique password, but whoever knows the password can read the pad. And I am pretty sure that if I asked IT about hosting an Etherpad instance for us, they’d laugh at me (fairly enough – it’s not the easiest thing in the world to maintain).
3. Request Tracker. Track each interview as a ticket, which preserves email exchanges with interview subjects for future researchers on the team. Use RTFMs (premade text snippets for quickly composing common replies) to standardize and semi-automate responses. We can assign tickets (interviews) to specific projects and set per-project and per-ticket permissions, and our interviewees only see the email-based interface which makes it easier for them. IT knows RT (I’m pretty sure they use it themselves), and we need something very close to plain vanilla, so they might even do it on our behalf. This is my favorite option so far, but it’s pretty… culture-shock inducing for non-software people (which is everyone else in my research group).

Ideas?

Anyone who writes and publishes (researchers, authors, students, and those who want to work with them) should read this little document on Author’s Rights [pdf] by Charles Bailey that gives a clear how-to and why-to introduction on getting started with open access. It covers reasons for making things open access, explains copyright and publisher agreements, shows you how to quickly find (and respond to) the policies of the publishers you use, explains creative commons licences and open access journals… I found it to be an easy-to-read guide (10 pages not counting cover page and bibliography) that I’ll use as a quick-reference in the future.

For anyone who wants to give a quick talk on open access, use or remix Isaac Gilman’s sharp little (public-domain!) slidedeck on “open access in 15 minutes or less.”

Finally, if you’re an open access advocate at your institution and looking for some empirical data to take to people, print out this handy crib sheet for conversations on open access with both authors and administrators. It contains a list of selling points for each audience, plus (I love this) links to actual data and examples to back up as many points as possible.

It doesn’t quite cover everything, though, so you’ll also want to download all 17 pages of The Open Access Citation Advantage by Alma Swan, which is an annotated bibliography covering (as of 2010) the empirical research that’s been done on the effects of OA on, quite simply, “the things researchers get evaluated on.” It’s essentially a big table of papers and their findings, so you can (for instance) search for studies specific to your field, sort by sample size, check their analytical approaches, and so on. They consistently find that, yes, Virginia, there is an OA citation advantage. (Of the 31 studies covered, 27 demonstrated a clear OA advantage, and 4 demonstrated either no advantage of a disadvantage — if anyone takes a look at those 4, I’d love to hear what you find!)

By open access, we mean its immediate, free availability on the public internet, permitting any users to read, download, copy, distribute, print, search or link to the full text of these articles, crawl them for indexing, pass  them as data to software or use them for any other lawful purpose.

–from Citation Advantage of Open Access Legal Scholarship

I spent it (unexpectedly) engrossed in Citation Advantage of Open Access Legal Scholarship (by James M. Donovan and Carol A. Watson), which took articles from an 18-year span from each of 3 law journals published by the University of Georgia and checked for correlations between citations of an article and its open access status (having “open access” status here was defined by “we could find the full text by using Google” – regardless of having the full text out there was technically legal or not).

Short version: yes, there was a nontrivial impact; OA papers got cited 58% more than non-OA ones. No surprise here; we’ve found this before. Two new things caught my attention, though.

First, how long does the “OA impact” last? 17 years. Donovan and Watson found the impact on citations trailed off over time, petering out at the 17 year mark. Why 17? No idea. (Would they have found different things if they’d looked at more than 18 years’ worth of data? I don’t know.) I wonder what you could do with a knowledge of that 17-year timeout – is that 17 years after publication, or 17 years after it first gets placed online? If it’s the latter, could individual researchers use this as a strategy to revitalize interest in work they did over two decades ago?

Second, OA had no (significant) impact on nonscholarly use of the material. The paper was written about legal scholarship, so what this means in practice is that while OA increases the use of an individual article by legal academics, it did not increase its use by courts. I wonder if similar things hold true in engineering; does OA make research more likely to be used by practicing engineers building real products, or does it only have an impact on engineering researchers? (Yes, I realize most practicing engineers hardly ever read research papers.) If it doesn’t, what strategies would facilitate the transfer of engineering research discoveries into the “real world” of “actual products” and the things that “working engineers” know?

Fascinating.

Next week I’ll finish the reading portion – far more efficiently, because I plan on coming in with printed copies of pre-triaged papers.  Boy, am I glad we budgeted some start-up time; this morning’s biggest accomplishment was getting a basic Zotero workflow up and running (not the software so much as the habits I developed in order to use it effectively). I only managed to read one paper, mostly because I spent far too much time taking overly detailed notes on it; in the future, I’ll reserve detailed note-taking for really important papers and write short summaries for the rest. (Next week I’ll arrive with a printed stack of papers already triaged by importance and really try to blast through them.)

I also found the first reading for the open* reading group Seb Benthall and I are going to do this term: selections from The Access Principle, which (as its subtitle says) makes a “Case for Open Access to Research and Scholarship” – and an empirical one, too.

You want clean laundry. You can put a washer and dryer in your own apartment and have access to it whenever you want, but you have to pay for the machines, maintain them, and are limited to one washer and one dryer load at all times unless you buy more machines. But why would you? That single machine is probably already sitting empty most of the time, because most people do a couple loads once a week.

But here’s the thing – you wanted clean laundry, not a washer and dryer. And all your neighbors probably want clean laundry as well. And furthermore, you’re likely to do laundry at different times; if you do your laundry on Saturday and Mr. Jones next door does his on Sundays, you and Mr. Jones could theoretically share the same washer and dryer and save money.

Scale that idea up into a business and you get the laundromat. It’s got a lot of machines, so whenever you need laundry done, you can walk over, stuff them into unused machines, and pay by the load.  You’re no longer limited to one washer and one dryer; if your little cousin gets the flu and projectile-vomits all over the house, you can do 10 loads simultaneously, and you don’t need to ask for permission to do it – you just pay for more loads. If you go visit your grandparents for a month, your laundry costs you $0. Not only do you not need to maintain the washers and dryers, you’re likely to be using bigger, better, and newer machines than you’d get for yourself.

The laundromat is like a public cloud.

Okay, said our parents. But then why would anyone want a private cloud?

Think about printers in your office, I said. Having your own washer-dryer is like giving everyone in the company their own desktop printer; using the copy shop across the street is like the laundromat. But as the big boss, you might not want all your employees using the copy shop across the street because… gee, that’s a lot of confidential information floating in a public place (whether your nervousness is justified or not). So what do you do? You get a giant super-printer, stick it in the office, and hook everybody up to that. That big printer is yours; it’s in your office, on your network, maintained by your IT staff. The printer is like a private cloud.

And in practice, you’re going to get a few employees with desktop printers, a bunch of big office printers that get heavy usage, and people running across the street to the copy shop when they’re on business trips or in a hurry to get 10,000 binders made right before the giant Sales meeting. That’s a hybrid model.

They got it.

From the introduction:

This is a preliminary and extremely incomplete literature review surveying the current academic scholarship on open source and education. It can be summarized in three words: there’s not much. I’ll begin… by discussing what I am not looking at… I am not looking at the use of open source software in educational contexts… open source as an IT solution. Nor am I looking at open source with a focus on content licensing… open source as an information access solution.

Instead, I am concerned with an analysis of what sorts of practices and processes for learning are exhibited in open source communities themselves and how these practices might be made transferable back to the classroom. In other words, I see open source as a way of operating learning communities: radically cross-functional, collaboratively constructed realtime transparency.

Aaaaand here’s the whole dang thing, below. Why push now? Well, after a certain point, I decided to stop agonizing over it and release early and often because hey, maybe what I need are some other eyeballs to kick me out of the mental ruts I’ve gotten into. Also, I finished writing it and turned it in 20 minutes ago (which may explain why the quality of the writing degrades considerably in the last 3 paragraphs). It’s been a rough week. Month. Semester, really.

Literature review: the open source way and education (alpha)

Essentially, I spent my entire first semester of grad school finding that nobody’s really done research on my topic of choice (engineering education happening in open source communities) before – which is both awesome (because that means I get to do it) and depressing (because WHY AM I ALL BY MYSELF OUT HERE I DON’T KNOW WHAT I’M DOING). And now I’m standing here unsure of what to do next; I want to read some of these more deeply, talk about them with people, have folks tell me where I’m wrong and why, get into debates over a bunch of these ideas, throw them off the wall, bounce them off other people’s heads…

And so I fling this out to you, my friends. Thoughts? I apologize that this is not in comic book form – it was a little harder to get away with that for a lit review… but I’ll keep trying.

I’m tremendously grateful to everyone who took a chance on me and helped me grow; if OLPC was where I took my first stumbles into the big, wide world of open source and learned to walk, Red Hat was where I learned to run, and to run alongside the best. The chance to grow both Fedora Marketing and FUDCon into mature, community-run entities, the chance to lead POSSE from an idea to a reality that’s touched dozens of professors and hundreds of students around the world, the opportunity to play a crucial role in the early stages of Teaching Open Source, the chance to learn from everyone around me – these are experiences I couldn’t have had anywhere else, and things I could not have done without my kickass teammates, mentors, and friends. The list is far too long to include here, but suffice it to say it’s been the people who’ve made the journey as amazing as it’s been, and I have no plans to say goodbye.

I’m still going to be in many of the same spaces I was in before (especially Fedora, TOS, and opensource.com), so in practical terms, what this means is that I’m swapping a .com email address for an .edu one so I can devote more time to graduate school and other personal projects along the same general theme of Awesomeness and Learning. And I’m still doing research on the open source way — what are the elements of the methodology, and how can they be made systematically and measurably transferable and reproducible? It looks like answering that question (or whatever it turns into) will be a large part of the next few years of my life, and I’m looking forward to it and will definitely keep you all updated.

It’s a big wide world out there. Time for the next adventure.

Our topic is how liberal arts students can get involved in open source.

Take a Walk in the Commons: Open Source and the Liberal Arts

As educators in the liberal arts, we try to prepare our students for a lifetime of learning. When we introduce our students to open communities as part of their classroom experience, we provide them with an opportunity to engage with the world around them and help solve real problems for real people. Participation in open communities as part of the undergraduate experience provides ways of developing and practicing communication, teamwork, and leadership skills—and contrary to popular belief, sometimes the most valuable contributions are the nontechnical ones.

In this talk, we’ll explore the wide range of opportunities for students to get involved in open source and open communities, with examples, case studies, and concrete next steps you can take as an educator or student to bring these opportunties into your classrooms and projects. Regardless of whether your interests are intellectual property law, computing in the sciences, enabling government transparency, language learning, or something else entirely, there’s an open community waiting for students to contribute and make a difference.

As of today (November 30, 2011) our slides are almost-but-not-quite done – comments welcome on what we do have, which is up on github. Slides were created with Beamer. Which brings me to noting two things I learned today:

Installing Beamer on Fedora 16

Beamer is a lovely little tool that uses LaTeX for making presentation slides. And all the Fedora installation instructions I found for Beamer are incorrect, so I’ll save you the hour of head-scratching I endured this afternoon and tell you that you need two packages beyond whatever comes with the default F16 install. Just yum install texlive-texmf-latex texinfo-tex and you should be all set to jump into the directory with beamer slides and make away.

If you want to try out beamer, here’s a quick little walkthrough that’ll get you looking at our current slides in 5 lines, assuming you already have git installed (if not, yum install git first).


git clone git@github.com:mchua/A-Walk-in-the-Commons.git # grab the code
cd A-Walk-in-the-Commons/presentation # go to the slide files
sudo yum install texlive-texmf-latex texinfo-tex # install beamer
make # builds the beamer slides into a pdf
evince a-walk-in-the-commons.pdf # and you've got them!


Changing the origin of your git repository

My git repository of the slides is actually a fork of Matt’s. I cloned his original, made and committed my changes, and then realized that I didn’t have push access to his repository, so I forked my own. But my commits were still lined up to push to Matt’s repository – how do I tell git to push to my new repository instead? It took a couple minutes of scanning through manpages and searching the intarwebs, so here’s the answer for posterity (for the record, the solution came when I finally realized we did this for SoaS documentation git issues).


git remote rm origin # remove the original repository (in this case, Matt's)
git remote add origin git@github.com:username/your-new-repository # add my new repository
git push origin master # and now it pushes to the right place