Recent Tweets
join our mailing list
* indicates required

About Ambleside

Ambleside Logic is led by Aaron Rosenbaum. Father of 3, Programming since 7, DevOps since 11 (hacking RSTS), exIngres, exCTP, exCohera. Sold two companies to Oracle, one to HP. Research + Strategy for NoSQL/BigData ecosystem implementors, vendors and investors.


Top 12 Oracle 12c Features - notes from AskTom's talk

Thom Kyte gave an entertaining, as usual, talk yesterday on the top 12 features in 12c.  Don't know if I'll make it to tommorrows talk (tough schedule placement after Wednesday night!).  Here are the highlights along with my snarky comments

#1 embedded PL/SQL - I applaud the uses in the simple cases Thom showed, I shudder to think about how this will be abused.

#2 improved defaults - Given all the new functionality, will this make things less bewildering for beginners? Don't think so.

#3 increased size limits for datatypes (32K) - Creating ever more oppourtunity to slow things down instead of using the proper datatypes and indexes.

#4 Top-N and pagination queries - Yeah!! Now if only the caches work properly when users do screwy things.

#5 row pattern matching (time series use case explained) - Now were talking.  This is incredible cool.

#6 Adaptive Execution plans (switching join types dynamically) - Cool - but why wasn't this done already?

#7 Enhanced Stats (query planning) - Proof is in the queries - didn't get enough detail.

#8 Temp Tables not generating redo/undo - Duh!

#9 Data Temp/Tiering - Probably my favorite feature of all

#10 App Level Continuity - Is this new?

#11 Data Redaction - Seems to make sense but will folks really use this vs. app code?

#12 Single Instance of Oracle running - need to see footprint...until then, no bigger claim than server only runs ESXi....

All and all, a strong release -much more focus on the ops guys than new business functionality but with EMC and VMware competing with Oracle for share of wallet with more strength than SAP, thats to be expected.


12c displaces large amounts of EMC Fast VP for RAC functionality

Exadata already has been displacing use of EMC VMAX in some accounts.  With Oracle's committments, although many features have lagged, having a true 2nd source for storage to run mission-critical apps on is just too critical not to support.  I think some of the technology has finally caught up to the strategic benefits.

Before 12c, there were important advtanges VMAX could offer vs. relying only on Exadata storage.  Those seem to be going away.

Here are the key advantages EMC has previously laid out:


  • Performance Tuning by allocating resources by policy
  • Better utilization and performance by hot/warm/cold auto-tiering
  • Simplified Management - virtual provisioning
  • Better Flexibility through pooled storage
  • Single pane of glass management


12c has obliterated all of these advantages.  

Performance tuning at the plug-in/container level exposes another level of granularity - or two - and allows for cross-system dependencies to be modeled.

The tiering policies in VMAX have never been that accurate - it's an overlay on another tiering strategy inside of Oracle already.  Extending ILM will be simpler to manage and should save a lot of space.

Single Pane management - It never really was single pane - DBA's still lived in Enterprise Manager.  Now DBA's can have more insight into the storage.

I think the real question for Oracle customers is whether they can stomach some of the ZFS pain, particulary those that were saved by EMC in the last decade after being burned by Sun.  Some of the hosting options will make that a little more palatable to pilot but some of my contacts say, yes, they are willing to give Oracle another chance.



VMWare DB Consolidation Projects - Oracle 12c just cut you off

VMware has had several efforts towards monetizing the Oracle DB Consolidation business.  Using vCenter has been relatively successful.  vFabric Data Director is a little more nascent - and I think will become more so now that its part of the suite.  VMWare sales folks really don't know how to sell to DBA's or Storage Admins.

VMwares effort has just come to a big stop sign.  12c container databases really bring the single pane administration that Data Director promised, but in a much more real way.  While Data Director took on admin rights, some light monitoring, cloning - and resource management, 12c goes a lot farther.

Consolidating resource management across all tiers - single DataGuard workload was the prime example - will reduce the workload of DBA's and Storage Admins - something VMWare just can't do.  VMWare golden image strategy dealt with deploying test + dev DB's but never dealt with patch management  Oracle 12c does.  I never really got straightforward answers about interactions with Oracle 11g and balloon memory.  My personal tests showed some clear anomolies.  Those problems go away too.

The message around database consolidation in 12c isn't tightly linked to the app deployment story - thats a weakness, and I haven't seen how the non-core RDMS products are dealt with (Hyperion/Endeca/TimesTen/etc).

Do you think 12c will slow VMWare Oracle consolidation projects? If not, why not?


Getting ready for Oracle OpenWorld 2012

Looking forward to hearing about 12c database.  I've heard rumors we're going to see a new generation of Exadata with better options at both low-end and high-end.  I'm especially interested in seeing the follow-up to the cloud announcements made over the summer.  Watch this space.



Is Zookeeper becoming the standard cluster manager?

I have been intrigued lately about the growth of Zookeeper as a general purpose cluster manager.  Zookeeper started as a clone of Chubby, a google project to maintain distributed locks.  Zookeeper is the core metadata filesystem for Hadoop.  Zookeeper maintains information on cluster participants, services and status - it's a cluster manager.

There have been some other uses in the past – Neo4j and Rackspace – but things seem to be accelerating lately.

Netflix’s Curator project is quite interesting – particulary the Service Discovery work as more general purpose workflow coordination.

Twitter is using Mesos on top of Zookeeper for meta data but also to instantiate services.

More influencial is Storm.  Built on top of Mesos.  Nathan's presentation at Strata 2012 was jammed.

LinkedIn wrote Norbert last year – seems very light weight but they are doing something similar to what Twitter is doing.

The most popular session about Zookeeper at 2012 Hadoop summit?

Dynamic Reconfiguration of Apache Zookeeper

Zookeeper is definitely not just for map reduce jobs anymore...