Well, it is early Wednesday morning here at the Westin hotel in Denver where the RMOUG Training Days conference is being held. I can’t sleep anyway so I thought I would write-up some of my impressions of yesterday’s presentations.
I appreciate all the effort people put in making their presentations. Since I have done Toastmasters I’ve learned to appreciate more what goes into being an effective speaker. But, the nature of my work is that I have to be critical of everything people say about technology. Maybe I should say “I have to think critically” instead of “be critical”. The problem with the type of work we do is that it involves a lot of money and that inevitably obscures the truth about the technical details of how things work. So, I want to just sit back and applaud but my technical side wants to tear apart every detail.
A nice perk of being a RMOUG presenter is that I got to attend the pre-conference workshops for free as well as the rest of the talks. In past conferences that I have spoken at that was not the case. So, I went to a four-hour Snowflake workshop. I have read a fair amount on Snowflake so much that the speaker presented was familiar. I wonder how people who had no Snowflake background perceived the talk? Being a nuts and bolts Oracle person I would have liked to dig in more to Snowflake internals and discuss its limitations. Surely any tool has things it does better and things that it does not do so well because of the choices that the developers made in its design. I’m interested in how Snowflake automatically partitions data across files on S3 and caches data in SSD and RAM at the compute level. At least, that is what the information on the web site suggests. But with cloud computing it seems that people frown upon looking under the covers. The goal is to spin up new systems quickly and Snowflake is fantastic at that. Also, it seems to get great performance with little effort. No tuning required! Anyway, it was a good presentation but didn’t get into nuts and bolts tuning and limitations which I would have liked to see.
I spent the rest of the day attending hour-long presentations on various topics. AWS offered a 3 hour session on setting up Oracle on RDS but since I’ve played with RDS at work I decided to skip it. Instead I went to mostly cloud and Devops sessions. I accidentally went to an Oracle performance session which was amusing. It was about tuning table scans in the cloud. The speaker claimed that in Oracle’s cloud you get sub-millisecond I/O which raised a bunch of questions in my mind. But the session was more about using Oracle database features to speed up a data warehouse query. It was fun but not what I expected.
I was really surprised by the Devops sessions. Apparently Oracle has some free Devops tools in their cloud that you can use for on premise work. My office is working with a variety of similar tools already so it is not something we would likely use. But it could be helpful to someone who doesn’t want to install the tools yourself. I’m hopeful that today’s Devops session(s) will fill in more details about how people are using Devlops with databases. I’m mostly interested in how to work with large amounts of data in Devops. It’s easy to store PL/SQL code in Git for versioning and push it out with Flywaydb or something like it. It is hard to make changes to large tables and have a good backout. Data seems to be Devops’s Achilles heel and I haven’t seen something that handles it well. I would love to hear about companies that have had success handling data changes with Devops tools.
Well, I’ve had one cup of coffee and Starbucks doesn’t open for another half hour but this is probably enough of a pre-dawn RMOUG data dump. Both of my talks are tomorrow so today is another day as a spectator. Likely it will be another day of cloud and Devops but I might sneak an Oracle performance talk in for one session.