This year, even though Oracle made huge announcements about 12c, I cared much more about non-database technology and spent most of my time researching desktop virtualization, solaris 11 and systems. So except for Larry’s keynote, I had not heard details about 12c all week. So I set out to ask people at the annual blogger meetup about their single most exciting feature and let them explain what it does. I was afraid that they’d all name the same one or two features but at the end of the event there were only very few duplicates and I had learned a lot from my peers. I may have gotten a few of these explanations wrong and I did not do much additional research (mostly because not much documentation or articles are available yet), so do not rely on the information being correct, rather think of it as how I first perceived them.
Philipe Fierens (@pfierens) was fast to reply with Data Guard Far Sync. A new proxy or repeater sits between both source and target database and let’s you use synchronous mode even over long distances with high latencies. This means you won’t have to choose between data availability and performance because you can get both even when the databases are far apart.
Jason Arneil (@jarneil) loves Cluster ASM for the fact that it will be possible to have multiple ASM instances running on the same system. The concern is that when you consolidate databases they all become dependant on only one ASM instance. And if that one has issues, all databases on the system will get taken down with it. Apparently, the problem is big enough that people were already fiddling with this in 11g and managed to start multiple ASMs even though it is not supported.
(update 2013/04/14: blogged about it)
Pluggable databases is definitely a high profile feature and Larry dedicated quite some time during his keynote to explain it. No wonder Øyvind Isene (@OyvindIsene), Frits Hoogland (@fritshoogland), Danny Bryant (@dbcapoeira) and Alex Gorbachev (@alexgorbachev) went with this one. Obviously the ability to run more than one database with just one instance is critical for consolidation projects. Previously you could consolidate many databases into one instance with many schemas but that could lead to issues with too little isolation and namespace conflicts. Running multiple instances is the other option but having redundant processes and memory regions is a waste of resources and hard to manage from a performance perspective. I also sat in Tom Kyte’s presentation on Wednesday and the way he described the concept behing PDB looked very similar to Solaris zone virtualization. Gunnar Bjarnason is skeptical that this will actually work well and already nicknamed the feature “possible database”. And James Morle (@JamesMorle) noticed a blooper in Larry’s keynote where he was talking about “plausible databases”. I have not yet looked at the replay but it must be true since this was actually picked up by the experts at informationweek.
There will be a new version of database console called Enterprise Manager Express that will ship out of the box and run on the local Oracle Server but provide much of the features that previously required an EM installation. Kellyn Pot’Vin (@DBAKevlar) and Doug Burns (@orcldoug) are excited about this.
Leighton Nelson (@leight0nn) and Gustavo Rene Antunez (@AntunezRene) are happy they may no longer have to restore a whole tablespace or even database when just a single table is broken because 12c promises to enhance RMAN in such a way that a single table point in time recovery will be possible.
Jeremiah Wilton (@oradebug) will be pleased to run multiple database replay sessions concurrently with Consolidated Database Replays.
Tim Hall (@oraclebase) will play with PL/SQL White Lists which allow explicitly specifying which packages or procedures may call another package. I am not sure if I understand this correctly but I think this is useful to not expose helper functions that are really just not meant to ever be called directly.
Fuad Arshad (@fuadar) will take advantage of global services which allows clients to connect to and load balance between any database that is running a certain service. Two-Way replication can be done by a mechanism of your choice. Sounds like this can fit the gap between RAC Clusters and desaster recovery replication scenarios.
Chet Justice (@oraclenerd) had no idea what his favourite feature was but he made sure to brag about our swim in the bay on Monday morning (blog with pictures by Kent Graziano). Hopefully, we can make this a regular annual thing and get even more people out there.
SQL improvements of regular expressions and row pattern matching are favourites for Cary Millsap (@CaryMillsap) and Greg Rahn (@GregRahn). Row pattern matching is an extension to analytic functions and it allows you to view peaks and dips for data along a partition-by clause. James Morle (@JamesMorle) thinks that simple SQL additions like row limits or for top-n queries are important since the existing workarounds are tedious and make working with Oracle feel outdated in contrast to mySQL or postgres.
While not strictly a new feature, Carlos Sierra (@csierra_usa) made sure I mention his passion for sqltxplain, a tool that will gather and save all information it can find about a given SQL statement for analysis by Oracle Support or yourself.
Jeremy Schneider (@jer_s) is intrigued by the promise of Application Continuity of masking planned or unplanned node failures by replaying sql transparantly if neccessary. I have spoken about the current situation of setting up applictions for high availability and would also like to see how app continuity can improve the handling of node outages.
Lisa Dobson (@lj_dobson) loves the improvements to partitioning. From the little I have heard about this it seems like there are a lot of smaller improvements that are just tying up loose ends and some bigger ones like a mechanism to automatically compress or move partitions based on their age.
Christo Kutrovsky (@kutrovsky) does not even care much about new features as long as old and known bugs are fixed. So when 12c becomes available he will first check if the limitations for parallel query on interval partitions are fixed.
Feature-slut Gwen Shapira (@gwenshap) (“I have a new one every week”) is happy to see 32k byte support for the varchar2 datatype because this reduces the effort and cost associated with migrations from other rdbms systems that already support larger character datatypes. Christo warned that while the storage size is determined by the number of characters actually used, the buffers on the client side must be able to support the defined size which may lead to problems there.
I also managed to talk to Tim Gorman (@timothyjgorman), Mogens Nørgaard (@mnorgaard), Roel Hartmann (@RoelH), Fahd Mirza (@fahdmirza) Alex Nuijten (@alexnuijten) and Ole Voss but did not get to do much nerd-talk. There was a contest about the most creative place on ones body to put a “love your data” stamp. Debra Lilley (@debralilley) took that one against Sheeri Cabral (@sheeri) and Ben Prusinski (@benprusinski) by placing a “Love your data stamp” on her leg. I don’t think it was the placement alone but rather the way she presented it. I guess whoever wants to be the keeper of Stanley (@StanleyACED) has to be a woman of many talents. Lillian Buziak from OTN and Pythian’s Paul Vallee (@paulvallee) sponsored the evening with fingerfood and drinks. And of course a huge thank you hug goes out to Vanessa Simmons (@pythiansimmons) for leaving it all on the ice and putting this event together.
Finally, I have not even mentioned Martin Nash (@martinpaulnash) and Karl Arao (@karlarao). I did not talk to them much at the event but we actually started Wednesday with a great run through San Francisco. In addition to a good workout, I also got some help from Martin by sharing some issues I am currently facing with an Oracle DB and being led in the right direction for fixing it. I just love Oracle OpenWorld.
What others are blogging about the meetup: