Thoughts about Oracle OpenWorld

I hate to admit it, but the only reason I went to Oracle OpenWorld was that none of the senior DBAs in the team wanted to go. The Boss agreed to send one DBA to the event, and if any of the more senior staff were interested, I’d have stayed at the office.

Fortunately, the older DBAs hated Oracle OpenWorld. They found it too crowded, too intense, too marketing oriented and not technical enough. And they are right, it is all this.

But to me, Oracle OpenWorld was pure magic.

I spend most of my days in a mundane DBA world – we export and import, we backup and recover, sometimes we get to tune some bad SQLs. After a year of this, and you get convinced that being a DBA is slightly boring.

And then, for one magical week, Oracle becomes exciting again. You meet people who are excited about what they do with Oracle – whether it is RAC, indexes, Model clause or ASM. What’s even better – in the Oracle Magic Kingdom, everything works. All the new features have no bugs, all patches install easily according to metalink instructions, Oracle support listens to your problems and has the right answers, you have sufficient maintenance windows to perform all your tasks, and performance issues are interesting and yet can be resolved within 45 minutes. Magic.

My favorite part was meeting people who are as excited about the database as I am. The unconference sessions and the blogger meetup was especially good for this. Most of the DBAs I meet on a regular basis view databases as their job, they think I’m rather strange and geeky for getting excited about new features and talking non-stop about storage options. Not to mention that my strange habit of playing with the database on weekends is considered as a symptom of a sever mental illness. In Oracle OpenWorld I got to meet a bunch of people who are every bit as geeky as I am, many are even more so. Saying “Wow! I’ll play with this feature on the weekend” did not cause a single person to take pity on my lack of social life.

I hope my excitement about OOW didn’t cause some of my senior co-workers to decide to go next year, because I want to go again.

Last day at Oracle OpenWorld

10:00 – RAC Internals – Fascinating session. The speakers covered some of the mysterious clusterware processes, explained how RAC manages group membership, how it protects us from the horrible split brain (I wish they had more details about IO fencing), what it OCR and how to manage it, and finally VIP explained! Did you know that when a node fails and the VIP moves to a different machine, the entire purpose of the exercise is to drop connections faster? That the VIP on the new node can’t accept connections? That was entirely new to me.

11:30 – Direct NFS – There were 3 speakers sharing the session, but Kevin Closson is by far the best of the three. He is simply an amazing speaker, probably the best I’ve heard in OOW. So, why use DNFS instead of traditional NFS?  It is easier to manage and configure, you can run Oracle with NFS on Windows (no idea why anyone would do that, but Kevin thought it was rather cool), and if you have multiple NICs for your NFS, DNFS will scale far better than the old “teaming”.

I’ve had some spare time before my next session, and I spent it by buying few books in the Oracle Bookstore. Two RAC books (because I’ve obviously didn’t satisfy my need for RAC information in this conference), one PL/SQL book and one “Best Software Writing” book for some light reading.

2:00 – Laurent Schneider’s unconference session about SQL Model –  SQL Model statement is a bit complicated, I’ve looked into it several times in the past, only to promptly disregard it as a nice idea but something too complicated to use in real applications.
With Laurent explaining it, SQL Model suddenly becomes accessible and practical. As something I may use when generating my next report and that will make things easier rather than more difficult. The session was really that good. He even posted the examples he used in his blog, so everyone can review it.

Oracle OpenWorld – Day 3

9:45 – Current Trends in Database Performance – A great start for what will be a great day. Andrew Holdsworth gave a fascinating session with many practical take aways. I wish I could summarize his talk here, but it would take me an hour and several pages, so maybe I’ll discuss it a bit in a later post.

11:15 – Best Practices for DBAs – I generally dislike the idea of best practices, and would have skipped the session except that I liked Arup Nanda’s articles in OTN and wanted to attend his session. It was a mistake. While Mr. Nanda is an energetic and engaging speaker, I still have a problem when some random practices of differing qualities are being promoted as being the best.

3:00 – RAC Performance Management – It would have been a terrific 3 hour session, but as a one hour session it simply went way too fast for me to understand. I got the part about diagnosing interconnect problems, but when they got to GC contention, congestion and latency – I simply could not follow. I wish Oracle would offer a one day seminar on the topic. I would love to attend that.

4:00 – Lutz Hartman’s unconference session about ASM – Made me wonder if migrating to ASM is really that simple, or perhaps it only seems so when an expert like Lutz is doing this.

Oracle OpenWorld – Day 2

Of course, with all the excitement of Oracle OpenWorld, I’ve had no time to edit and post my notes during the conference itself, so you are getting everything in a big batch now.

12:15 – RAC SIG expert panel – It was supposed to be a great opportunity to ask RAC questions and have the very people who wrote the product answer. In reality it was a big disappointment. I didn’t get to ask my cool questions (mostly about load balancing and fail over mechanisms) while other people asked questions that were completely unrelated to RAC, like how to move datafiles. I could see that even the experts were rather disappointed.
I compensated for the disappointment by strolling down to the demo booths later and asking my questions there. I also bought two RAC books, just to be extra sure.

2:00 – 11g & Compliance – Dan Morgan and Victoria Whitlock scheduled an unconference session about compliance with SOX and the likes in 11g. Due to some quirk of luck, I turned out to be the only attendant of this wonderful session. Everyone else will have to travel all the way to Seattle to learn how 11g makes SOX compliance that much easier. Dan Morgan told me about the wonderful Audit Vault – which was just what we needed for a big auditing project and we didn’t hear of it before, and also how to use 11g’s PL/SQL warnings to automatically detect many issues with code before I upload it to production. This was by far the most useful session I’ve attended this day.

15:15 – DataPump in 11g – Datapump is one of the best features in 10g, and it got even better in 11g. My favorite improvement is the inline compression of the data dump and the ability to import without uncompressing first. Other nice features are inline encryption, the ability to remap data during the import while keeping certain properties of the data format (for example, generate bogus email addresses for the test system), and the ability to convert LOBs to SecureFiles.

16:45 – Top 11 new features in 11g – This talk was not as excited as I expected. Probably because I’ve already spent few hours hearing about new features in 11g, and also because such a rich set of features deserved more than one hour. 5 minutes per feature are not enough. A nice discovery was that Oracle added pivot operator to SQL, I was not aware of this before the lecture. No more ugly case statements when you need to turn your tables sideways. All the other features were the usual suspects – cache, partitions, real application testing, data pump, compression, yadda, yadda.

The highlight of the day was actually the evening. I went to the blogger meetup, and in addition to drinking excellent beer on Oracle’s expense, I also got to meet the wonderful people that I’ve been writing my morning news for the last year. I have to update my blog roll now because I’ve met few cool people who write great blogs that I haven’t been following up on. It was a great experience to know that Oracle Bloggers are not only smart, creative and great writers, they are also very nice people and its really fun to hang out with them.

One of the things that are kind of sad to me now that OpenWorld is over is that I’ve met all these great people, but they all live in the UK, Australia, Switzerland, Oregon, Canada and what not. There is no way I’m going to meet any of them again for a beer and a chat for the next year, at least. Thats a really depressing thought. As much as I enjoy the virtual connection, talking to people over a glass of beer is much much better.

OOW2K7 – Day 1

11:00 – GRID Design  – I arrived late to this session due to traffic delays, but I still managed to get most of it. It was interesting, but not exciting. The speaker reminded us to pay close attention to throughput and not just to size, and to design for flexibility. He stormed through everything from buses to application design, but never got any details about any of them, just some general good advice.

1:15 – DBA 2.0 in the No Slide Zone – That was the big disappointment of the day.  the description in the content catalog seemed very interesting: “The next generation DBA will play a more strategic role in managing increasing numbers of complex and business-critical database systems. Learn how to take automation, management and decision-making to the next level”, and I assumed that Tom Kyte will talk about how DBAs can  play a more strategic role in the organization.  This topic seemed interesting enough that I decided to skip Steven Feuerstein’s SQL Addiction talk and attend the DBA 2.0 session.
Unfortunately, the topic was in fact Enterprise Manager, in which I have very little interest, and Tom Kyte didn’t speak, he just played host. I felt significant physical pain as the audience laughed at the old DBA who was editing his scripts in Emacs and then applauded the new DBA who point and clicked his way through Enterprise Manager.
Shame on you Oracle, for the false advertisement.

2:00 – Richard Foote talks about Indexes at the Unconference. That was simply the best technical lecture I’ve heard all day, and maybe ever. Richard Foote is a very engaging speaker. He went over some facts and myths and really made everything much clearer to me. I’ll try to hear him tomorrow as well. It is so much fun to listen to someone who is actually passionate about what he is doing and doesn’t sound like an overtrained monkey.

3:15 – Storage Capacity Planning on the back of the envelope. This session was useful. The speaker briefly explained how to measure storage usage in an existing system and how to use this to estimate growth, and then went on to explain how to interpert the capacity capabilities of the hardware components – CPUs, drivers, devices, FC cards and routers, arrays and disks. Taking all this into account requires a rather large envelope, but it was a good session anyway.

4:30 – 11g Performance features. This was the most exciting session, simply because Oracle added many neat features in 11g. I never imagined that table compression could actually improve performance, I didn’t pay enough attention to secure files and they sound very exciting too, combine this with new cache capabilities, partitioning features and new advisors and all of the sudden I’m actually excited about 11g instead of the usual fear that production DBAs feel toward new versions.

This was a great day. My mind is about to blow from all the new information and this is only the first day of the conference. I hope that tomorrow I’ll be a bit less shell shocked.

My Oracle OpenWorld 2007 schedule

It is tough building a schedule – seems like all the presentations I want to attend are at the exact same time. And then half of the “unconference” sessions are not even scheduled yet. Anyway, here is what I came up with:

Sunday – I’m still debating whether or not I’ll attend Larry Ellison’s keynote. On one hand, everyone says it will be great. On the other hand, I’m not staying in SF, so it means another commute from the valley, and I’m not sure Larry Ellison is worth two hours commute.

Monday –

11:00 AM – Grid Design with Oracle RAC
1:15 PM – DBA 2.0 (This clashes with Steven Feuerstein’s talk about SQL Addiction, which I also really want to attend.)
2:00 PM – Mark Hurd keynote
3:15 PM – Back-of-the-Envelope Database Storage Design
4:45 PM – Oracle Database 11g: Next-Generation Performance and Scalability (This one clashes with an Executive Solution Session about IT Governance that sounds interesting).

Tuesday –

10:45 AM – I got automatically registered to some “Executive Solution Session” about virtualization. I’ve no idea what it is, but I’m planning to find out.
12:15 PM – Oracle RAC SIG Expert Panel
3:15 PM – Oracle Database 11g Data Pump: Data Compression, Encryption, and More
4:45 PM – The Top 10–No, 11–New Features of Oracle Database 11g
7:30 PM – The Oracle Bloggers’ meeting : the Thirsty Bear Brewing Co, 661 Howard St (Thanks APC for pointing out this gem!)

Wednesday –

8:30 AM – Keynote, Jonathan Schwartz, Sun Microsystems Inc.
9:45 AM – Current Trends in Database Performance
11:15 AM – IOUG: DBA Best Practices from the Field (I hate Best Practices, but I like Arup Nanda’s articles)
12:45 AM – Keynotes, Michael Dell and Larry Ellison. : Moscone North – Hall D
3:00 PM – Practical Performance Management for Oracle Real Application Clusters (Yes, I really am attending every RAC related session in the conference)

Thursday –

10:00 AM –Demystifying Oracle RAC Internals
11:30 AM – Improve the Performance and Manageability of Your NFS Environment with Oracle Database 11g Direct NFS Client (Yes! )

Ooh! I’m going to OOW!

Just like every DBA, next week I’ll be at the Oracle Open World convention up in San Francisco.

Unlike most DBAs, I’m much more anxious about it. Well, its my first time in a convention of any sort, so I’m having a difficult time choosing between all the events – with so many presentations, keynotes, unconference meetings, no slide talks, and parties, I’ve no idea what I should be doing and where to turn my attention. I’d say that I feel like a kid in a candy store, but this is so much more intense.

What’s more, there are a bunch of people I’d like to meet – bloggers that I like reading, and the nice people who read my blog. I’ve no idea how to go about it, though. There are supposed to be some 50,000 people at the conference, and I have about 10 readers, so that totally rules out random meetings. Arranging a meeting seems a bit pretentious though. But I’m more worried about meeting the bloggers that I love reading. I mean, I can probably go up to Kevin Closson after his lecture and tell him that I love his blog, but isn’t it a bit too much of a fan boy thing?

And then there is the network thing. Everyone is telling me that the most important part in the conference is networking. Thats great, but I’ve no clue how to go about it. Should I keep talking to random people and give them my business card, or is there more scientific approach to it?

Worrying all the time and over planning everything are great traits for a production DBA. But few days before my first conference, I wish I could have been a bit more like that developer I’m working with, the one who didn’t register to any session and is planning to just show up one morning and wing it. Conferences really shouldn’t be so damn complicated.