I’d like to share my personal impressions of this years UK Oracle User Group Conference. They are of course purely subjective and by no means comprehensive. First things first: It was a great event, in my opinion! Fantastic speakers including many Oracle Celebrities, very nice location and service by the UKOUG. Especially I liked the opportunity to meet so many guys in person that I communicated with previously via social media 🙂
Every single presentation I attended was worth it, to say the least. My goal in the following list of presentations is to highlight just one thing about them that I think is worthwhile to memorize, although it was of course not the only important thing in that talk.
On Monday, I saw Julian Dyke with ‚Is RAT Worth Catching?‘ Since I have taught Real Application Testing in many 11g New Features & Performance Tuning courses, I was curious what he had to say about it. Although he appeared not very enthusiastic about it – the price is of course a concern here – a key point here was: Database Replay is indeed good for capture & replay of realistic production workload and will show the impact of a change pretty accurately measured in DB Time.
Next was Alex Gorbachev with ‚Oracle ASM New Features‘. It was about 12c, so everything here is under Oracle’s safe harbor statement:
‚All information provided outlines our general product direction. It’s intended for information purposes only, and may not be incorporated into any contract. It is not a commitment to deliver any material, code or functionality, and should not be relied upon in making a purchasing decision. The development, release, and timing of any features or functionality described for Oracle’s products remains at the sole discretion of Oracle.‘ Having said that, one major thing I found interesting was the potential upcoming of clusters where not every node needs an ASM instance running, thereby reducing the need of ASM instance communication in spite of large clusters with very many nodes.
Another highlight was Christian Antognini with ‚Shareable Cursors‘. Especially I liked his many live demonstrations. Very instructive! One take away: Starting from 22.214.171.124, a parent cursor that has more than 100 child cursors is obsoleted, which is controlled by _cursor_obsolete_threshold
On Tuesday I saw Emre Baransel with ‚Oracle Data Guard Deep Dive‘. I knew the information here before, but I wanted to see him present it after we have chatted so often via twitter already 🙂 One of many things to memorize: There is no point in using ARCH Redo Transport any more for Max Performance because ASYNC has also no performance impact but delivers far better protection than ARCH.
Next came Joel Goodman with ‚Global Resource Management in RAC‘. One key point out of many important things my colleague said: The resource mastership of an Oracle block is determined by an internal algorithm but can be dynamically remastered (in case of blocks on the segment level) so that another node becomes the master of a certain segment when blocks out of that segment are frequently required from that node.
James Morle was referring about ‚Building a Winning Oracle Database Architecture‘. One remarkable thing in my opinion here: It is probably much less painful to follow the mainstream (regarding Oracle Database Architecture) even if that is boring and uncool. Better be boring & reliable instead of cool & chaotic – now doesn’t that pretty much reflect the image of a DBA? 🙂
Larry Carpenter was then talking about ‚Oracle Data Guard Zero Data Loss Protection at Any Distance‘ Again that is 12c stuff and falls under the same safe harbor disclaimer as above. Larry is one of the very few people I have an Oracle book from in the shelf and probably the best source of knowledge about Data Guard you can find. Looks like there will be something called ‚Far Sync‘ standby databases in the future that receive redo in SYNC mode, cascading it with ASYNC (not with archive shipping) to potentially extremely distant remote standby databases.
Then came Tom Kyte with ‚What’s New in Security in the Latest Generation of Database Technology‘. Again see safe harbor disclaimer above, as it is 12c. We will probably have a standard procedure that analyzes the actually used privileges of a given user, making it easy to avoid granting too powerful or too many privileges was one cool thing here.
On Wednesday, I attended Dan Norris with ‚Exadata X3: The Fourth Generation‘. Although I knew that presentation, I was curious to see him in person, after we have had some conversation about Exadata before. One key point here: The Write-Back Cache stores writes mirrored across cells and persistent across cell reboot. I mention that here because I have seen some misconceptions especially about this in the Exadata community.
Then came Andy Colvin with ‚Exadata Zero Downtime Migration‘. It turned out that it was not exactly zero but about 2 minutes downtime, because they did it with Data Guard switchover. Particular striking in my opinion: Even with primary databases of multi-TB sizes, duplicate from active database can reasonably be done – it may take days to complete but who cares as long as there is no downtime during that on the primary 🙂
Finally, I saw Tanel Poder with an amazing presentation about ‚Troubleshooting the Most Complex Performance Issues I’ve Seen‘. Very instructive indeed! One thing to memorize out of many: If a wait-event is not properly instrumented by Oracle development, it may falsely show up as CPU-Wait. You could drill down with pstack then to reveal the root cause.