Comments for Conor O'Mahony's Database Diary

Comment on IBM DB2 Strikes Another Blow to Oracle Database by Paulo Matos

Paulo Matos — Fri, 11 May 2012 13:55:03 +0000

Did you consider using socket instead of cores/processors in your calculations?

A thing essential to me is an HADR mechanism, at least having a coherent stand-by database, which in DB2 is offered (included) right from entry editions and Oracle has only on Enterprise Edition at an extra cost.

Most of my clients’ needs are fulfilled with workgroup/standard edition and this feature request in Oracle makes the price too high.

Comment on Baltic Bank Moves from Oracle to DB2 to Improve Performance, Lower Costs, and Increase Availability by Thiago Rossetto

Thiago Rossetto — Thu, 03 May 2012 03:50:45 +0000

Reblogged this on Thiago Rossetto and commented: Oracle Database on Sun servers to IBM DB2 on Power Systems servers. Is there anything related to DB2 for z/OS?

Comment on Announcing DB2 10 & InfoSphere Warehouse 10 by Announcing DB2 10 & InfoSphere Warehouse 10 | Alex Rosa – Database Specialist

Tue, 03 Apr 2012 17:52:01 +0000

[...] http://database-diary.com/2012/04/03/announcing-db2-10-infosphere-warehouse-10/ -22.382931 -46.946498 Share this:FacebookTwitterLike this:LikeBe the first to like this post. [...]

Comment on Oracle's Shameful Study: Oracle Database 11gR2 vs. IBM DB2 9.7 by The Choice of a New Generation « Julian Dontcheff's Database Blog

The Choice of a New Generation « Julian Dontcheff's Database Blog — Sat, 31 Mar 2012 08:44:41 +0000

[...] interesting, in my opinion, is Conor O’Mahony’s analysis of that study entitled Oracle’s Shameful Study: Oracle Database 11gR2 vs. IBM DB2 9.7. Scott Hayes [...]

Comment on Need Help Determining Hadoop Split Sizes? Use Adaptive MapReduce Instead! by Bert

Bert — Sat, 31 Mar 2012 05:58:55 +0000

“Adaptive Map/Reduce” is truly adaptive. In general, one map task might have a lot more work to do than others, even if all map tasks get the same amount of input data. This might be true because there is skew in the input data. A map task with lots of work will not grab more work/splits, while a map task with little work will grab more work/splits to work on.

The CombineInputFormat is static and no dynamic adjustment of work across map tasks happens.

Comment on Need Help Determining Hadoop Split Sizes? Use Adaptive MapReduce Instead! by hadooper

hadooper — Fri, 30 Mar 2012 18:02:58 +0000

This is what CombineInputFormat of Hadoop also does.

Comment on Oracle Reduce their Exadata Projections by Kevin Closson

Kevin Closson — Tue, 28 Feb 2012 22:56:41 +0000

Thanks for the kind words, Conor. You and I share a bit of IBM heritage (Sequent acquisition).

I too would like to see someone lend credence to the use of the term “units.” Considering the offerings range from quarter to full rack the term is too nebulous to have value. The quote from the transcript is “In Q3, we plan to sell over 300 Exadata and Exalogic engineered systems.” I do believe they’ll hit that number because Exalogic sells well and clumping the more disruptive Exadata into the less disruptive Exalogic numbers makes for simpler smoke and mirrors.

Comment on Oracle Reduce their Exadata Projections by Conor O'Mahony

Conor O'Mahony — Tue, 28 Feb 2012 22:07:41 +0000

Hi Kevin, Thanks so much for your comment. I am a long-time fan of your writings.

Does anyone reading this blog post happen to know what constitutes a “unit” for Exadata? Does a “unit” correspond to an actual physical shipping system? Or does a “unit” correspond to something else?

Comment on Oracle Reduce their Exadata Projections by kevinclosson

kevinclosson — Tue, 28 Feb 2012 21:41:49 +0000

In the first week of Oct 2011 at OOW (more than half way through Q2FY12) the projections were a supposed 3,000 units “this year” : http://youtu.be/jAmgVbuFZwY We know now that only 200 units moved in Q2FY12 so I wonder what all that 3,000 stuff was about?

There just a little more than a full day left in this leap year and to the end of Q3FY12. We’ll see about the 300.

Comment on IBM DB2 Strikes Another Blow to Oracle Database by Conor O'Mahony

Conor O'Mahony — Mon, 27 Feb 2012 13:57:11 +0000

Hi Dirk,

Regarding your other claims, I believe that the information in this blog post is accurate. Let me explain why…

Compression: Oracle base compression is included in Oracle Enterprise Edition. However, that’s only compression for bulk load. If you want to get compression for inserts (like the DB2 Deep Compression) then you need Oracle Advanced Compression.

Disaster Recovery: To get functionality equivalent to what HADR provides, you have to buy Active Data Guard. For instance, with HADR you can do reads against the standby server. With Oracle Database, when you do a read on standby, you can’t apply logs while reading at the same time (like we do with HADR) unless you purchase Active Data Guard.

SQL Developer: I imagine the “ease of use” of the two products is a matter of opinion.

Advanced Security: DB2′s Advanced Access Control is similar to Oracle Label Security. DB2′s Advanced Access Control is multi-level security (for classified data).

Statspack: Yes, Statspack is free… just like DB2 snapshot monitors are free in all editions. But Optim Performance Manager provides added value to make performance monitoring and tuning easier. Just like AWR and ADDM (the underpinnings of Enterprise Manager Diagnostic pack) make things easier (where you have to pay for it).

Golden Gate: Q-Replication provides replication bi-direction much like Golden Gate. With Data Guard it’s one way, with Q-Replication it’s peer-to-peer. With Data Guard it’s the full database, with Q-Replication or Golden Gate you can replicate a subset of the tables.

People may not need all of these add-ons. However, when you consider that if you do need a handful of them, the price for Oracle is going to grow very quickly when compared to the price for DB2 Advanced Server Enterprise Edition (which includes them all).

Regards,
Conor.