HotSos, Concurrency, Papers and Related Thoughts

So I finally finished my “Seven Sins of Concurrency” paper. The one I’m going to present at HotSos Symposium next month.

Writing a paper was a very educating experience. In the few weekends I’ve been working on it, I learned more about Oracle than I did during the last year. Its been tiring and exhilarating. Of course I learned all sorts of new things about concurrency, consistency, locks and latches, but I also learned about merge, autonomous transactions and packages.

I used to think that writing a paper is a bit like writing a technical blog post. There are some similarities, but the big difference is that blogging is fun, while writing a paper goes beyond fun into painful. Kind of like riding my bike. I do it for fun, but when I train for a race, it is pure pain. Of course, this pain is what leads to improvements. Still, I would need a long time to recover from this.

Writing a paper and publishing it is also very scary. The paper contains some things that were not done before, otherwise it will not be worth while. But since they were never done before, there is some chance that I got things completely wrong. Since I published all my scripts and data, if I’m wrong, someone can find out how wrong I am. While I love corrections and always want to improve, the idea of being wrong this publicly scares me a lot. I’m putting a lot of myself out there for the world to criticize. Of course, going this far out of one’s comfort zone is also critical for improvement. But I can understand presenters who give those lists of tips and best practices – it seems less scary.

While thinking about the paper, and reading some other papers for inspiration, I noticed three interesting things that separate the papers I see in the Oracle world and the papers I see from the academia:

  1. Academic papers are usually written by more than one person, which is quite rare for Oracle papers.
  2. Academic papers always reference the latest and greatest work of their peers. Oracle papers, when they reference at all, usually reference older and more established work. Academics always present their work as improvement on something their peers did, while in the Oracle world we try to present our papers as original, even when it is not.
  3. Academic papers are published in peer reviewed journals. Oracle papers are rarely published and rarely peer reviewed.

I think that point #3 is the cause of the previous points. Since we don’t have peer reviewed journals with their associated prestige, all the prestige in Oracle world belongs to conferences and presentations. We write papers for conferences, and since we plan to be the one presenting on stage, we don’t collaborate. Of course, since there are no journals, it is more difficult to reference and improve on the most recent work done in our field. Just keeping updated is challenging.

It looks like a peer reviewed journal could greatly improve the quality of scientific Oracle papers out there. Of course, peer reviewed journal would be expensive and may not be justifiable from business perspective. Maybe we should start with a peer reviewed blog. I even have a title – “Oak Table Chronicles” 🙂

Advertisements

2 Comments on “HotSos, Concurrency, Papers and Related Thoughts”

  1. Rob van Wijk says:

    Congratulations with the paper. It was a very interesting read.

    I’ve emailed some minor details. No major flaws 🙂

  2. Iggy Fernandez says:

    Hi, Chen,

    Most presentations at user conferences and trade shows like Oracle OpenWorld don’t rise to the level of academic research. For example, my last presentation was “Xtreme SQL Tuning; The Tuning Limbo”—a blend of entertainment and instruction calculated to please a crowd.

    There are definitely plenty of academically rigorous papers featuring Oracle but they are probably only found at the academic conferences. The ACM website lists lots of papers referencing Oracle and other database technologies. For example, see http://portal.acm.org/toc.cfm?id=1385269 (Proceedings of the 1st international workshop on Testing database systems).

    I’m sure there are exceptions to the rule that academically rigorous papers are not found at user conferences and tradeshows. One really good paper that comes to mind is Juan Loaiza’s paper “Optimal Storage Configuration Made Easy” which introduced the SAME methodology—Stripe and Mirror Everything. (http://www.oracle.com/technology/deploy/availability/pdf/oow2000_same.pdf)

    Iggy


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s