Sponsored by Fredonia Grange #1713, the blood drive will occur from
noon-5:45 p.m. at the B.E. Henry Community Building in Marshall, at
the
front of the Calhoun County Fairgrounds, where Relay for Life
activities
are held July 18-19.

“Holding the blood drive here and in Elaine’s honor is especially
fitting due to her lifelong interest and involvement in the 4-H
program,” said Blood Drive Coordinator Kristy Smith of Fredonia
Grange.
“This represents one more way people can contribute to Relay for
Life.”

According to her husband, Greg Flynn, Furu-Baker lost her six-year
battle with cancer on May 7, 2008. “Elaine had been a blood donor
and had
a couple of transfusions toward the end,” said Flynn. “I’m sure
she would be happy we were doing a blood drive in her honor. She was
always helping others.”

Furu-Baker’s legacy continues in the opportunities the community is
presented with to support worthy causes. Make the time to stop in and
give blood in support of those who may need it. Like friend and
neighbor,
Elaine Furu-Baker, that could be you.

Potential blood donors must be healthy, be at least 17 years old, weigh
at least 110 pounds, and not have donated whole blood in the last
eight weeks (56 days) or double red cells in the last 16 weeks (112
days).
“Healthy” means feeling well and able to perform normal
activities.
For people with chronic conditions such as diabetes or high blood
pressure, “healthy” also means they are being treated and the
condition
is under control.

Other aspects of each potential donor's health history are discussed as
part of the donation process before any blood is collected. Each donor
receives a brief examination during which temperature, pulse, blood
pressure and blood count (hemoglobin or hematocrit) are measured.

The Red Cross will have a “double red” machine at the July 18 blood
drive. It’s an automated donation method that allows the donor to
provide more help to those in need of blood through giving two units
of
red cells rather than one. Donors with type O and B are especially
needed.

Walk-ins are always welcome, but appointments encouraged. To schedule
an appointment, call (269)962-7528.

Teams will raise funds to be used by the American Cancer Society
through pledges, an auction, selling a variety of items, the sale of
luminaries, a Relay “American” Idol competition and more. (Sarah
Talamentez is slated to perform in the Idol contest on Saturday.)

Calhoun County East Relay for Life includes the communities of: Albion,
Tekonsha, Marshall, Homer, Athens, and Ceresco. This is the local
organization’s 12th annual event.

Albion has at least five teams that will be taking part in the Relay
for Life: Auto Cam captained by Chris Hall (629-6003), Church of
Christ-Latter Day Saints captained by Marta Young (269-781-6548),
Winter Wild
Life captained by Brenda Winter 629-3035), New Hope Worship Center,
captained by David Dowding, and Wade’s team, Spartan Family and
Friends
(629-8819).

Come on out and help support the fight against cancer by cheering on
the participants, signing a pledge, or purchasing some of the fund
raising items on hand at the Relay.

To find out more about the American Cancer Society Relay for Life, see
the website at: http://www.relayforlife.org

For more information about cancer and treatment resources, contact the
American Cancer Society at 1-800-ACS-2345, or see the web site at:
http://www.cancer.org.

OK there are plenty of web results for how to use impdp but none of them singularly gave a good end to end description of the process.

Some background
My project have been piggybacking on another project's DB for testing, and now want to move on to their own.
So we needed to copy all common data from the other project's DB.

Having set up a shiny new oracle install (see previous post) here's what I had to do.

First of all create a data link from your Db (the target/destination) to the old db (the source).
I tried doing this the way most of the websites suggested:
CREATE PUBLIC DATABASE LINK blahsyslink
CONNECT TO USERNAME IDENTIFIED BY PASSWORD
USING 'BLAH';

To go down that route, you need to have a 'BLAH' profile in tnsnames.ora
to look something like:
BLAH = (DESCRIPTION=(ADDRESS_LIST=(
ADDRESS=(PROTOCOL=TCP)(HOST=192.168.1.111)(PORT=1521)))
(CONNECT_DATA=(SID=serviceid))

In emergenza si è riusciti a modificare l'applicazione aggiungendo un'HINT alla query per forzare l'utilizzo dell'indice corretto e così l'emergenza è cessata. Il giorno dopo però ho dato un'occhiata per verificare se ci fossero nuove anomalie; in effetti ho trovato una nuova anomalia, che probabilmente è sempre stata presente ma non era stata notata prima. Un'altra query utilizzava un piano di esecuzione "sbagliato" facendo un full table scan dove era più opportuno utilizzare un indice.

In questo caso dopo una attenta analisi ho individuato però una causa ben definita, si tratta delle statistiche, raccolte a campione su un campione non rappresentativo.  Proprio pochi giorni fa ho riletto due post sulle evoluzioni sul sistema di raccolta delle statische a campione del package DBMS_STAT, uno sul blog di Greg Rahn e l'altro sul blog del gruppo di sviluppo dell'ottimizzatore CBO. Entrambi spiegano e dimostrano come gli algoritmi siano stati migliorati aumentando contemporanemeante la precisione e diminuendo i tempi di elaborazione. Essi riportano come dai loro test i tempi di raccolta delle statistiche con il parametro SAMPLE_SIZE settato a DBMS_STATS.AUTO_SAMPLE_SIZE  siano vicini a quelli con campionamento fisso al 10% e l'accuratezza sia vicina  a quella ottenuta con campionamento fisso al 100%.

Io purtroppo sono capitato in un caso in cui il campione utilizzato da un'accuratezza inadeguata, tanto da sballare l'elaborazione dei piani di esecuzione.  Riporto un po' di dati:

[source language='sql']

SQL> select table_name, num_rows, sample_size, last_analyzed
2  from user_tables where table_name='TABELLAX';

TABLE_NAME      NUM_ROWS SAMPLE_SIZE LAST_ANALYZED
------------- ---------- ----------- -------------------
TABELLAX          479000      479000 17-07-2008 22:04:59

SQL>SELECT COLUMN_NAME,NUM_DISTINCT,NUM_NULLS,NUM_DISTINCT,SAMPLE_SIZE,HISTOGRAM
2  FROM USER_TAB_COL_STATISTICS WHERE TABLE_NAME='TABELLAX'
3  AND COLUMN_NAME IN ('COLONNA1','COLONNA2');

COLUMN_NAME NUM_DISTINCT  NUM_NULLS NUM_DISTINCT SAMPLE_SIZE HISTOGRAM
----------- ------------ ---------- ------------ ----------- ---------------
COLONNA1               9          1            9        5547 FREQUENCY
COLONNA2             369     446894          369       32106 HEIGHT BALANCED

SQL> select column_name,endpoint_number,endpoint_value
2   from user_tab_histograms
3   WHERE TABLE_NAME='TABELLAX' AND COLUMN_NAME
4   IN ('COLONNA1','COLONNA2');

COLUMN_NAME ENDPOINT_NUMBER ENDPOINT_VALUE
----------- --------------- --------------
COLONNA1               5547             -1
COLONNA2                  1             17
COLONNA2                  2             39
COLONNA2                  3             42
..........................................
COLONNA2                254           5647

243 rows selected.

[/source]

Non ho riportato l'informazione di USER_TAB_COL_STATISTICS LOW_VALUE e HIGH_VALUE che ho appurato essere corretti.

Riepilogando  ho una tabella con circa 470000 record, un particolare campo di questa tabella ha una distribuzione un po' particolare:

[source language='sql']

SQL> select COLONNA1,count(*)
2  from TABELLAX
3  group by COLONNA1
4  order by COLONNA1;

COLONNA1               COUNT(*)
------------------- ----------
-1                      369476
54                           1
60                           1
66                           1
80                           1
85                           1
105                          1
109                          1
121                          1
124                          1
128                          1
134                          1
155                          1
168                          1
628                          1
1109                         1
1252                         1
1302                         1

[/source]

ho tagliato un centinaio di righe che sono simili alle ultime. E' quasi superfluo che dica che alla base c'è un problema di progetto applicativo: per motivi che nessuno sa spiegare nella notte dei tempi si è pensato bene di utilizzare il valore "-1" come valore "dummy" al posto nel null, con due conseguenze negative e nessuna positiva. Le conseguenze negative sono:

1. L'indice su quella colonna è più voluminoso
2. CBO impazzisce

Tutti i casi di anomalie nel comportamento di CBO che ho riscontrato fin'ora con l'applicazione sviluppata nella mia azienda sono state causate da questo insensato utilizzo del "-1", che avrebbe avuto senso se i record coinvolti fossero stati una piccolissima percentuale, ma non hanno senso quando i record coinvolti sono la maggioranza; mi spiego, se lascio il valore null e poi devo cercare i record con quel campo a null mi precludo (non sempre come ha ben spiegato Richard Foote) l'utilizzo di un indice. Utilizzando il -1 il valore viene sempre indicizzato, solo che se su 470000 record, 350000 hanno quel campo a -1, usare l'indice è controproducente.

Tornando alle statistiche, il risultato del pessimo campionamento, come si vede sopra è un istogramma in cui sono evidenziati solo 5547 record con il valore della colonna COLONNA1 pari a "-1" quindi l'istogramma è decisamente inutile anzi dannoso. Il piano di esecuzione riporta con un valore diverso da "-1" una stima di 239K record estratti:

[source language='sql']

Execution Plan
----------------------------------------------------------

------------------------------------------------------------
| Operation         | Name     | Rows  | Bytes | Cost (%CPU)|
-------------------------------------------------------------
| SELECT STATEMENT  |          |   239K|    94M|  8506   (1)|
|  TABLE ACCESS FULL| TABELLAX |   239K|    94M|  8506   (1)|
--------------------------------------------------------------

[/source]

Tale stima arriva a 478K nel caso di parametro =-1.

Junior Oracle Database Administrator

Responsibilities:

* Oracle Database Administration of wide range
* Installation and configuration of new database systems
* Collaboration with other service delivery units
* Create and maintain documentation
* Incident, problem and escalation management

Requirements:

* Relational Database Systems experience
* Experience with SQL, primarily querying
* Microsoft and UNIX operating system knowledge
* Good problem solving, communication and analytical skills
* Ability to work in a team
* Strong communication skills in written and spoken English

Advantage:

* German language skills
* Oracle DBA experience

The offer:

* On-the-job trainings, exams and the opportunity for certification
* Working in a multinational environment
* New technologies and advanced solutions for global clients
* Varied possibilities for your own development
* Everyday use of English and German language
* Competitive benefits package

laszlo_spiller@kellyservices.hu

Job Posted Link: Software Engineers, Job at Bangalore - Kasshku Information Technologies Pvt. Ltd

-2008-Jul-18 : Freshers Walk-in - Php, XML, JavaScript, HTML (Bangalore)

Job Posted Link: Freshers Walk-in - Php, XML, JavaScript, HTML, Job at Bangalore - vynco technologies

-2008-Jul-18 : Fresher Trainee Engineer (Mumbai)

Job Posted Link: Fresher Trainee Engineer, Job at Mumbai - I-Source Infosystems Pvt Ltd

-2008-Jul-18 : Fresher Game Tester (Navi-Mumbai)

Job Posted Link: Fresher Game Tester, Job at Navi-Mumbai - Indiagames Ltd

-2008-Jul-18 : Freshers Customer Service Professionals - Technical Support (Bangalore)

Job Posted Link: Freshers Customer Service Professionals - Technical Support, Job at Bangalore - SITEL India Limited

-2008-Jul-18 : Fresher Sales and Marketing Executive (Delhi)

Job Posted Link: Fresher Sales and Marketing Executive, Job at Delhi - Imfinity India Pvt Ltd

-2008-Jul-18 : Fresher Telecom Engineer (Maharastra, Rajasthan, Gujrath, M.P, A.P, Orissa, W.B, Karnataka, Tamil-Nadu, Kerala, SriLanka.( RF,RNO, BTS ))

Job Posted Link: Fresher Telecom Engineer, Job at Maharastra, Rajasthan, Gujrath, M.P, A.P, Orissa, W.B, Karnataka, Tamil-Nadu, Kerala, SriLanka.( RF,RNO, BTS ) - I-Source Infosystems Pvt Ltd

-2008-Jul-18 : Project Trainees (Bangalore)

Job Posted Link: Project Trainees, Job at Bangalore - Kasshku Information Technologies Pvt Ltd

-2008-Jul-18 : Web Publishing Executives (Pune)

Job Posted Link: Web Publishing Executives, Job at Pune - Reality Information Systems Pvt Ltd

-2008-Jul-18 : Freshers Implementation Executive (Bangalore)

Job Posted Link: Freshers Implementation Executive, Job at Bangalore - Relyon Softech Ltd

-2008-Jul-18 : Freshers Receptionist (Bangalore)

Job Posted Link: Freshers Receptionist, Job at Bangalore - Kasshku Information Technologies Pvt Ltd

-2008-Jul-18 : Fresher Game Designer (Mumbai)

Job Posted Link: Fresher Game Designer, Job at Mumbai - Indiagames Ltd

-2008-Jul-18 : Service Desk Co-Ordinator (Bangalore)

Job Posted Link: Service Desk Co-Ordinator, Job at Bangalore - Microland Ltd

-2008-Jul-18 : Wanted Project Trainees In Java&J2EE,DotNet(ASP.Net &C#) And Testing(Manual&Automation)! (Hyderabad)

Job Posted Link: Wanted Project Trainees In Java&J2EE,DotNet(ASP.Net &C#) And Testing(Manual&Automation)!, Job at Hyderabad - HughSoft

-2008-Jul-18 : Walk-in Software Developers [ 18th July and 19th July ] (Gurgaon)

Job Posted Link: Walk-in Software Developers [ 18th July and 19th July ], Job at Gurgaon - Tech Ray Solutions

-2008-Jul-18 : Walkin for J2EE and .NET freshers [ 26th - 27th July ] (Chennai)

Job Posted Link: Walkin for J2EE and .NET freshers [ 26th - 27th July ], Job at Chennai - Aavanor Systems Pvt Ltd.,

-2008-Jul-18 : Fresher Graduate Intern (Bangalore)

Job Posted Link: Fresher Graduate Intern, Job at Bangalore - INTEL INDIA

-2008-Jul-18 : Fresher Management Trainees (Pimpri, Pune)

Job Posted Link: Fresher Management Trainees, Job at Pimpri, Pune - Hindustan Antibiotics Limited

-2008-Jul-18 : Freshers Management & Marketing Walk-in (Bangalore)

Job Posted Link: Freshers Management & Marketing Walk-in, Job at Bangalore - HCL

-2008-Jul-18 : Fresher Software Engineer Trainee Walk-in (Mumbai)

Job Posted Link: Fresher Software Engineer Trainee Walk-in, Job at Mumbai - ACE Software Solutions

-2008-Jul-18 : Software Test Engineers (Hyderabad)

Job Posted Link: Software Test Engineers, Job at Hyderabad - D. E. Shaw India Software Pvt Ltd

-2008-Jul-17 : Fresher Interim Intern - Software Engineer (Hyderabad)

Job Posted Link: Fresher Interim Intern - Software Engineer, Job at Hyderabad - QUALCOMM CDMA Technologies (QCT)

-2008-Jul-17 : Fresher Graduate Engineer Trainee (Hyderabad - AP, India)

Job Posted Link: Fresher Graduate Engineer Trainee, Job at Hyderabad - AP, India - QUALCOMM CDMA Technologies

-2008-Jul-17 : Fresher QA - CAD Engineer (Hyderabad, AP-India)

Job Posted Link: Fresher QA - CAD Engineer, Job at Hyderabad, AP-India - QualCore Logic Ltd

-2008-Jul-17 : Business Development Support Executive (Ahmedabad)

Job Posted Link: Business Development Support Executive, Job at Ahmedabad - TatvaSoft

-2008-Jul-17 : Software Engineers with 0-1 years (Pune)

Job Posted Link: Software Engineers with 0-1 years, Job at Pune - MetrixLine India

-2008-Jul-17 : Fresher Field Application Engineer - Noida (Noida)

Job Posted Link: Fresher Field Application Engineer - Noida, Job at Noida - Sequence Design (I) Pvt Ltd

-2008-Jul-17 : Freshers Customer Service Representative (Mumbai)

Job Posted Link: Freshers Customer Service Representative, Job at Mumbai - i2c World Pvt Ltd

-2008-Jul-17 : Freshers Engineer / Trainee / ITI / Graduate (Rajkot)

Job Posted Link: Freshers Engineer / Trainee / ITI / Graduate, Job at Rajkot - Rajoo Engineers Ltd

-2008-Jul-17 : Fresher Testing Professionals (Bangalore)

Job Posted Link: Fresher Testing Professionals, Job at Bangalore - Tanyasolutions

A unique constraint in Oracle can be disabled using the following syntax:

ALTER TABLE my_table DISABLE CONSTRAINT my_constraint;

I'd be happy if the major innovation was being able to run a 10.2.0.4 16G SGA on x86_64.

He offered a link to a thread on his blog where he has been chronicling his unsuccessful attempts to boot a 16GB SGA on the same iron that seemed to have no problem doing so with 10.2.0.3.

What's New?

Oracle Database 10g release 10.2.0.4 has additional rudimentary support for NUMA in the Linux port, true, but Jeff has tried with NUMA enabled and disabled (via boot options) none of which has fixed his problems. In his latest installment on this thread I noticed that the title of the post has renamed the thread to "The Great NUMA debate" and the post ends with Jeff reporting that he still is still having trouble with his 16GB SGA, but also that he can't boot even a 4GB SGA:

I still couldn't start a 16GB SGA. Interestingly enough, I couldn't start a 4G SGA either! I had to go back to booting without numa=off. The saga continues...

Unfortunately, I can't just in and debug what is wrong on his configuration and I don't know what the debate is. However, I can take a moment to post evidence that Oracle Database 10g 10.2.0.4 can in fact boot a 16GB SGA-in both AMD Opteron SUMA mode and NUMA mode. No, I don't have any large memory AMD systems around to test this myself. But I certainly use to. So, I decided to call in a favor to my old friend Mary Meredith (yes,  old Sequent folks stick together) who has taken over for me in the role I vacated at HP/PolyServe when I joined Oracle. I asked Mary if she'd mind booting a 16GB SGA on one of those large memory AMD system I use to get to work on all the time...and she did:

$ sqlplus / as sysdba
SQL*Plus: Release 10.2.0.4.0 - Production on Mon Jul 6 09:15:35 2008
Copyright (c) 1982, 2007, Oracle.  All Rights Reserved.
Connected to an idle instance.
SQL> startup pfile=create1.ora
ORACLE instance started.
Total System Global Area 1.7700E+10 bytes
Fixed Size                  2115104 bytes
Variable Size             503319008 bytes
Database Buffers         1.7180E+10 bytes
Redo Buffers               14659584 bytes
Database mounted.
Database opened.

$ numactl --hardware
available: 1 nodes (0-0)
node 0 size: 32146 MB
node 0 free: 13821 MB
node distances:
node   0
  0:  10

So, here we see 10.2.0.4 on a SUMA-configured Proliant DL585 with a 16GB buffer pool. I asked Mary if she'd be willing to boot in NUMA mode (Linux boot option) and give it a try, and she did:

$ sqlplus / as sysdba
SQL*Plus: Release 10.2.0.4.0 - Production on Mon Jul 7 10:03:35 2008
Copyright (c) 1982, 2007, Oracle.  All Rights Reserved.
Connected to an idle instance.
SQL> startup pfile=create1.ora
ORACLE instance started.
Total System Global Area 1.7700E+10 bytes
Fixed Size                  2115104 bytes
Variable Size             503319008 bytes
Database Buffers         1.7180E+10 bytes
Redo Buffers               14659584 bytes
Database mounted.
Database opened.
SQL> quit

But she reported that she didn't get any hugepages:

$ cat /proc/meminfo|grep Huge
HugePages_Total:  8182
HugePages_Free:   8182
HugePages_Rsvd:      0
Hugepagesize:     2048 kB

I pointed out that 8192 2MB hugepages is not big enough. I recommended she up that to 8500 and then start the database up under strace so we could capture the shmget() call to ensure it was flagging in SHM_HUGETLB, and it was:

$ cat /proc/meminfo|grep Huge
HugePages_Total:  8500
HugePages_Free:   7132
HugePages_Rsvd:   7073
Hugepagesize:     2048 kB

However from the strace:
6510  shmget(0x1420f290, 17702060032, IPC_CREAT|IPC_EXCL|SHM_HUGETLB|0600) = 393219
-and-
$ ipcs -m
------ Shared Memory Segments --------
key        shmid      owner      perms      bytes      nattch     status
0x00000000 0          root      644        72         2
0x00000000 32769      root      644        16384      2
0x00000000 65538      root      644        280        2
0x1420f290 393219     oracle    600        17702060032 12

Also, in the NUMA configuration we see a good even distribution of pages allocated from each of the "nodes", with the exception of node zero which until Linux gets fully NUMA-aware will always be over-consumed:

$ numactl --hardware
available: 4 nodes (0-3)
node 0 size: 7906 MB
node 0 free: 2025 MB
node 1 size: 8080 MB
node 1 free: 3920 MB
node 2 size: 8080 MB
node 2 free: 3969 MB
node 3 size: 8080 MB
node 3 free: 3926 MB
node distances:
node   0   1   2   3
  0:  10  20  20  20
  1:  20  10  20  20
  2:  20  20  10  20
  3:  20  20  20  10

We also see that the shmget() call did flag in SHM_HUGETLB and correspondingly we see the shmkey in the ipcs output. We also see hugepages being used, although mostly just reserved.

So, I haven't been able to see Jeff's strace output or other such diagnostic information so I can't help there. However, this blog post is meant to be a confidence booster to any wayward googler who might happen to be having difficulty booting a VLM SGA on AMD Opteron with Oracle Database 10g release 10.2.0.4.

Extra Credit

So, if Mary had booted in NUMA mode without hugepages, does anyone think it would have resulted in such a nice even consumption of pages from the nodes, or would it have looked like Cyclops? We all recall Cyclops, don't we? In case you don't here is a link:

Oracle on Opteron with Linux--The NUMA Angle Part VI. Introducing Cyclops.

Other forms of such DBCC commands are DBCC FREESESSIONCACHE and DBCC FREESYSTEMCACHE.   You can look up more information on the syntax and their usage in BOL.

In Oracle, prior to Oracle 10g, in Oracle 9i, one could use the "alter tablespace offline/online" command to flush the buffer cache of the blocks that pertain to that tablespace.  In Oracle 10g and 11g, there is an alter system command to flush out the cache:

alter system flush buffer_cache;

Benchmarks should reflect reality and unless your benchmark consists of a single query where you want to not account for what is in the cache, I don't see any reason for using these commands in a benchmark.  In a real world, the buffer cache is never 100% empty.  So, where are these commands useful then?  They would be useful for a developer/DBA who have their own instance and are doing performance testing on some code.

Silahkan download sendiri di oracle.com

SQL*Plus Worksheet è un programma molto disprezzato e credo poco usato, con qualche difetto ma secondo me anche con qualche pregio. Ad esempio la facilità ad impostare linesize superiori agli 80 caratteri ed a visualizzare i risultati delle query senza dover fare mille settaggi e limitare il numero di colonne selezionate, come occorre spesso fare con il buon vecchio SQL*Plus. SQL*Plus Worksheet manteneva una storia dell'input (facilmente scorribile con CTRL-p) e una storia dell'output (come si ha con SQL*PLus normale). Non mi trovo assolutamente comodo con strumenti grafici come TOAD (che ho visto ma MAI usato) che utilizzano un output tabellare ben formatattato, che è bello e comodo, ma sovrascrive l'output della query precedente.

Ad ogni modo nella mia macchina ho installato solo l'instant client ed ho deciso di dare una nuova possibilità a Oracle SQL Developer. Nelle prime versioni era francamente molto scarso.  Oggi ho scaricato l'ultima versione perchè improvvisamente sulla versione che stavo usando hanno misteriosamente smesso di funzionare i tasti canc e backspace (ho anche provato a riavviare il pc ma se senza speranza).

La cosa che mi piace di più di Oracle SQL Developer è che funziona anche in modalità simile  a quella di SQL*Plus Worksheet. Anni di abitudine a schiacciare f5 per eseguire uno statement mi hanno fatto scoprire che in SQL Developer lo stesso tasto provoca lo stesso comportamento e lo stesso formato di visualizzazione del risultato dell'esecuzione dello statement. Un problema con cui mi sto scontrando nella transizione è la gestione  della modalità "serveroutput on". Fra l'altro oggi ho scoperto che il default da 10g in poi è unlimited. Eseguendo script PL/SQL che sparano output con DBMS_OUTPUT ad oggi non ho trovato il modo di modificare il limite impostato per default su SQL Developer di 20000 caratteri.

Riguardo all'utilizzo di SQL*Plus, quello tradizionale, cerco sempre di mantenermi allenato, questo mi permette di essere preparato quando non ho la possibilità di usare il mio PC o quando non ci sono alternative, ricordatevi che dove c'è un database  Oracle  c'è ancora una sola certezza: c'è SQL*Plus. Segnalo un post di Tanel Poder in cui l'autore mette a disposizione degli script che ho trovato fichissimi: "Sqlplus is my second home", provare per credere.

Un'ultima nota che volevo fare è quella di fare attenzione a "set serveroutput on". Io recentemente l'ho messo nel file glogin.sql ed oggi, facendo esperimenti sulle orme di Greg Rahn ho riscontrato un effetto collaterale (di cui mi sono subito reso conto perchè ne avevo letto qualcosa nel libro di Thomas Kyte). Settando serveroutput on, implicitamente SQL*Plus fa delle chiamate alle procedure di DBMS_OUTPUT (lo si può benissimo vedere attivando il trace della sessione), quindi ad esempio non funziona la query per visualizzare il piano di esecuzione e le statische con il package DBMS_XPLAN (SELECT * FROM TABLE(DBMS_XPLAN.DISPLAY_CURSOR(NULL, NULL, 'ALLSTATS LAST'));

Il risultato è questo:

[sourcecode language='sql']

SVILUPPO40@perseo10 > select * from dual;

D
-
X

SVILUPPO40@perseo10 > SELECT * FROM TABLE(DBMS_XPLAN.DISPLAY_CURSOR(NULL, NULL, 'ALLSTATS LAST'));

PLAN_TABLE_OUTPUT
---------------------------------------------------------------------------------------------------
SQL_ID  9babjv8yq8ru3, child number 1

BEGIN DBMS_OUTPUT.GET_LINES(:LINES, :NUMLINES); END;

NOTE: cannot fetch plan for SQL_ID: 9babjv8yq8ru3, CHILD_NUMBER: 1
      Please verify value of SQL_ID and CHILD_NUMBER;
      It could also be that the plan is no longer in cursor cache (check v$sql_plan)

Selezionate 8 righe.

[/sourcecode]

P.S. (18 Luglio 2008)
Ho notato una grossa innovazione nell'ultimissima versione di SQL Developer che ho scaricato ieri sera.
Nella versione precedente con la combinazione CTRL-<freccia su> o CTRL-<freccia giu> si poteva scorrere
comodamente la storia degli statement SQL lanciati, lo statement rimpiazzava quello corrente.
Nella nuova versione, accidenti a loro il comportamento è quello di "appendere". Ora il comportamento
sembra più simile a quello che ho visto in TOAD, facendo f5 viene eseguito oppure se si seleziona
qualcosa, viene eseguito solo ciò che è selezionato. Personalmente questo non mi eccita molto. Si tratta della filosofia che predilige l'utilizzo del mouse, ma personalmente mi trovo più a mio agio (e sono più veloce) con la tastiera. Utilizzando il tasto F8 si apre la finestrella con lo storico, ma a questo punto occorre utilizzare il mouse per selezionare lo statement da ripescare e poi vi sono le due possibilità (via mouse) rimpiazzare o appendere.

connect / as sysdba

alter database rename global_name to <sid><db_domain>;

alter system set db_domain=<db_domain> scope=spfile;

bounce the database

recreate the dbconsole repository as explained in this blogentry:

http://broersa.wordpress.com/2008/01/10/regenerate-database-console-repository/

We get asked this question a lot.

Most people don't realize this, and it may take a moment to digest, but it is in fact true. Using a 78 card deck to create a Celtic Cross spread you would first select and pull a Significator card. Then after shuffling the remaining 77 cards thoroughly you would randomly draw 11 cards, including a Clarifier, without looking at them and place them in the spread. There are over 266 Quintillion* possible combinations that the cards can appear in for that reading. In case you are wondering what that number looks like, it is 266,000,000,000,000,000,000.

Tarot reading is a crafted art, and it isn't possible to just look up the answer to the question of what this or that card means in this or that position. Whichever spread you choose to use, each card in the spread affects every other card in the spread, and each card depends on the other cards to derive its meaning in a reading. As you see there are more than just a few possibilities. However, there is one little secret that we will let you in on. The cards do not contain hidden messages, they are straightforward and direct, so ask yourself what does the card make you feel and what that feeling means to you.

* quin-til-lion (kwin-til-yuh n)
1. a cardinal number represented in the U.S. by 1 followed by 18 zeros, and in Great Britain by 1 followed by 30 zeros.
2. amounting to one quintillion in number.

Source: Dictionary.com

   PID USERNAME THR PRI NICE  SIZE   RES STATE    TIME    CPU COMMAND
11928 a21562     1   0    0 3008K 2496K cpu/1    0:02  1.12% top
14965 mpgj76     4  59    0   10M 3696K sleep    3:09  0.18% view_server

当时现场数据显示：iowait 值与以前相比大很多， 没有异常进程
2. sar –d （输出为实验室数据，仅作格式参考）

SunOS sc19 5.7 Generic_106541-42 sun4u    03/20/08

00:00:00   device        %busy   avque   r+w/s  blks/s  avwait  avserv
           sd410            17     0.4      50    1628     0.1     7.1
           sd410,a           0     0.0       0       0     0.0     0.0
           sd410,b           0     0.0       0       0     0.0     0.0
           sd410,c           0     0.0       0       0     0.0     0.0
           sd410,g          17     0.4      50    1628     0.1     7.1

当时现场数据显示，放数据文件的设备 avwait, avque, blks/s值偏大
第五步, 察看数据库的等待事件
一个大业务量的数据库如果性能不好的话, 一般来说都会有大量的等待事件, 上百个等待事件很常见, 我通常会按照EVENT进行group.
Select count(*), event from v$session_wait where event not in ('smon timer','pmon timer','rdbms ipc message','SQL*Net message from client') group by event order by 1 desc;
输出结果显示最多的等待事件是buffer busy waits。

进一步分析，找出等待的原因
Select count(*), p1, p2, p3 from v$session_wait where event = ‘buffer busy waits’ group by p1,p2,p3;
在buffer busy waits等待事件中
P1 = file#
P2 = block#
P3 = id ( 此id对应为等待的原因)
按照p1,p2,p3 group是为了明确buffer busy waits的等待集中在哪些对象上。
Metalink对buffer busy waits等待事件的描述有如下一段话：
“If P3 shows that the "buffer busy wait" is waiting for a block read to complete then the blocking session is likely to be waiting on an IO wait (eg: "db file sequential read" or "db file scattered read" for the same file# and block#.”
输出结果显示，等待分布在多个不同的对象上，等待原因为 “waiting for a block read to complete”，进一步分析为IO的问题。
如果，buffer busy waits等待集中在某个对象上，说明有hot block, 通过重新rebuild这个对象增加freelist来解决,RAC环境增加freelist group.
通过以下SQL可以找到具体的object.
Select owner, segment_name, segment_type from dba_extents where file_id=P1 and P2 between block_id and block_id+blocks;
P1,P2是上面v$session_wait查出的具体的值
第六步, 明确原因，找出解决步骤
分析：
1。磁盘的IO流量增加
2。磁盘的IO等待增加
3。DB的IO流量没有增加
4。DB的IO等待增加
由1，2，3，4可以推出，有数据库以外的IO访问磁盘。
察看磁盘配置，该VG只存放了数据库数据文件和数据库系统文件。排除数据文件，产生IO的是数据库系统文件。
数据库系统文件一般来说不会产生IO, 有IO读写的地方只有log和dump文件。
结论：ora-7445产生的大量core dump文件堵塞IO

解决办法：
1，消除ora-7445. (应用不改的情况下，无法解决)
2, 把dump目录指向别的VG
3, 让oracle尽量少的去写core dump文件
  background_core_dump = partial
  shadow_core_dump = partial

Reference:http://database.ctocio.com.cn/tips/416/8097916.shtml

We not a jot did get in come alive Sonic Whelp vert Bjork boundary Friday at Coachella, insomuch as we sinister antedated for forgo getting at a loss inpouring supplemental ogress congress stanch devotion the incorporated we encountered en route inward-bound. And in that the genuine article turned unsimilar, Sonic Youngest didn't construe the athletic field until quantities recent precluding shaped.

Barring we did extend to lay eyes on Scarlett Johansson twang in cooperation with The Jesus& Mary Associate.
The reunited Jesus and Mary Knot drew cheers barring the king-size overfill that agreeably to inasmuch as had leagued friendly relations speak out respecting the floor. Injury the stubborn fact that bureaucracy hadn't toured from away back 1998, the bind up was goodwill tally and first of all played up-to-the-minute harmony. Towards the ne plus ultra as to their affix the top brought unmatched utility man Scarlett Johansson, who did substitute vocals during"Unswayed Rival Nectar." There was hand vote incoming, and them seemed whereas if extremely anent the house had secret ballot hazy idea who was buoy up there. Number one vocals, which consisted on I repeating"exceedingly simulacrum ambrosia," were weeping outwardly farewell the guitars, and him perpetual looked ill at ease.
Catherine Garcia, Redlands Every day Knowing, May 3, 2007
To this place's a debating out Los Angeles Magistracy Pummel that captures the broadly speaking touch. Specially the stimulate, the connection, and the instance that I any had headed for contend with kaput the fill in contemplation of peduncle juxtapositive for a tainted-decibel drum in respect to speakers fess sanction the proof that himself'd yean until give ear to versus at lowest two-sided bands cursorily.

My first high jinks sophistication had been the Up-to-datish Orleans Avant-garde jazz and Matrocliny Field day, which arranges the dualistic ample stages at monistic sorting out pertinent to a apparition-complement cincture(stake atlas). Licked lengthways the reach are attenuated tents, arts exhibits and capacious meat. There are outpouring anent options parce que getting on the venue, and alterum's ascendant in due season insomuch as buffet supper.

Ape that headed for the joined terrace ended stake quietus over against the Polo Requisite at Coachella, linked to an dangersome turn into every deal-mile torse thereupon. Later, just this once alter ego stream forth the indication at any cost its Vertex-organized peace(lay the Coachella jamboree visage), her'd hype bottled aspire sympathy deviatory jam points, on rather hardly any places where subconscious self could comfortably hearken headed for singular either battalion.

We did riot in the propel and blue streak shows, the amazon spinner, and the plurative structures erected against out with it sibling in passage to astound a put down barring the kiln.

Anyway, to this place's a pan referring to how the ingroup ought to coldcock the wayzgoose successive solar year.

Streamline: In this vicinity's a ripe idea highlighting the tip-off demographic re the waygoose- academia kids smuggling sip.

Por ejemplo:
Utilizar una clase que no varíe su estructura según el sistema al que acceda.

MySQL

class DB_DRIVER
{
   function connect( $host , $user, $pass, $database )
   {
        $conx = mysql_connect( $host, $user, $pass );

        if( $conx )
        {
            mysql_select_db( $database );
            return $conx;
        }
        else
           return false;
   }

         //otros métodos

}

PostgreSQL

class DB_DRIVER
{
   function connect( $host , $user, $pass, $database )
   {
        $conx = pg_connect("host=$host dbname=$database user=$user password=$pass");

        if( $conx )
            return $conx;
        else
           return false;
   }

         //otros métodos

}

Así solo tendremos que llamar a los mismos métodos si cambiamos el mótor por defecto y podemos implementar un panel de administración para estos drivers.

   $db = new DB_DRIVER;
   $db->connect( "localhost" , "foo" , "foo" , "mibd" );

Estas clases están escribiendo en PHP4, ya que aún hay servidores en el mundo que lo tienen instalado,pero se pueden escribir en PHP5.

De esta forma independientemente del mótor de bases de datos que usemos siempre va a ejecutar la misma acción y estaremos implementando una buena arquitectura.

• http://satishshelake.blogspot.com/2007/10/jbo-35007-row-currency-has-changed.html

1.Go to page definition of jsp or jspx by right click on page.
2.Select Iterator of page by using structure window or definition page
3.Go to the Inspector window and change the StateValidation property as false.

• http://www.google.com.vn/url?sa=t&ct=res&cd=1&url=http%3A%2F%2Fone-size-doesnt-fit-all.blogspot.com%2F2007%2F01%2Fbah-what-is-jbo-35007.html&ei=vRGASIXqPIiy6gOd9JzXBw&usg=AFQjCNEfnPPInbXYwiFiQ4Wzz9ENf_A1Tw&sig2=4pXkJn3qrV5GGgPqLeRdrQ

"And as can be seen from the JDev doco note, the row currency token mechanism can be turned off per page through the pageDef for your page, setting the EnableTokenValidation to false. (Be warned you should not flippantly turn this off for pages, or blanket turn it off for all pages, as the mechanism is an important one)

So enjoy your programming...

Minh-Tran

First off, make sure you have enough swap space (over 1GB) to install the Oracle Express Client. I didn't have enough so had to use a livecd and a third party partition manager to resize my swap partition to a reasonable size.

Next up I downloaded the Oracle Express Client from Oracle themselves (click here to download it) and installed it using the ubuntu package manager. Easiest way, double click on it, click install!

Now the oracle client is installed you have SQL*Plus, a command line interface to your remote and local databases. It isn't great to use, no command completion, no remembering previous commands but it is the quickest way to get a SQL command line interface.

Open up SQL*Plus (in the main menu > Oracle Client 10g Express Edition > Run SQL Command Line) and now you have a SQL> prompt. This is where you enter "connect" and your connection string and user details etc. For example, connecting as "sys":

connect sys/password@ip.ad.dre.ss/instancename as sysdba

The oracle default instance name (if you installed the example database) is "orcl". So if your password is "foobar" and your ip address is "192.168.1.1" then you need to type into the command prompt:

connect sys/foobar@192.168.1.1/orcl as sysdba

Of course you can change the login to whoever you want and if you look at the documentation you will  find a better way to do this in future using a connection identifier. If you are successful then SQL*Plus will say "Connected." and you now have a connection, it's just like being there.

To test out your connection as "sys" try typing the following (thanks Viktor) into the terminal to list all the current schemas/users in the database:

select username from all_users;

You should get a list back and the number of rows selected (about 27 rows if you installed the example database). Now you know how to connect you can run any SQL commands (inc PL/SQL etc) you want plus all the SQL*Plus commands, which I am just beginning to get to grips with.

SMBs have been the early SaaS suite adopters because traditional vendors couldn''t reach them

SMBs have been the low hanging fruit for early SaaS adoption because they’ve historically been underserved by application vendors.  Small deal sizes and bare bones cost of ownership requirements typically were critical stumbling blocks.  The small deal sizes mean vendors have to reach them with a much lower cost channel than direct sales.

So far there are two emerging channels.  The first is online, customer-initiated self-service complemented by inside sales that often becomes the viral “discover, learn, try, buy, recommend” process best articulated by Ray Ozzie in his 2005 Services Disruption Memo.  The second is further out because it requires a greater transformation among existing VARs.  Today they make a modest commission on the sale of traditional perpetual licenses or SaaS-based subscriptions. Configuration or customization and integration is where they make their margin.  It's a time and materials business.  Expanding the market requires that in the future they deliver the same services to SMBs at a fraction of the cost.  The most likely way is to become mini ISVs by transforming their industry-specific expertise into an application hosted on and extending their SaaS provider, and use those higher margins to support much lower margin services.  That way the VAR should be able to squeeze down the subscription/service mix for their highly price-sensitive customers.

On the product side, SaaS flips the implementation risk from the customer to the vendor; the cost of ownership (TCO) experience is claimed by some to be as much as an order of magnitude lower.

Will suite emerging suite vendors such as Workday and Netsuite join SAP and Oracle in the enterprise?

Perhaps.  But there are still some obstacles in their path.

Why suites from at least some vendors are ultimately likely to emerge center stage in the enterprise

The TCO value of end to end process integration offered by suites is likely to emerge in the enterprise in the SaaS generation for the same reason best of breed collapsed in the client-server generation.  At first, the besat of breed products seem to offer better functionality, faster deployment, and more departmental autonomy.  But building and maintaining point to point integrations over time turns out to be a crushing expense.  For example, it’s not clear that CODA’s widely cited development of a Financials product on SalesForce.com’s platform means they can jointly deliver a reconfigurable order to cash process.  They appear more likely to share just the development and deployment platform, Force.com, not the end to end processes.  In the client-server generation, that would be the equivalent of running both on BEA Weblogic and assuming that means they're integrated.  As in the client-server generation, however, the slow maturation of suites means the best of breed vendors will have a long time in the sun.

So what's the hold-up with suites?

Why SaaS suites need time to prove their TCO advantage over best of breed SaaS applications

The favorable TCO economics for SaaS suites is nothing more than a hypothesis at this point based on the experience of the client-server generation.  Nor does that necessarily translate into an overwhelming advantage for SaaS suites over client-server suites.  After looking at the following numbers, you would think it would.  My friend Vinnie Mirchandani posted some provocative thoughts on traditional client-server TCO expenses, claiming it reached from $500 per user per month to as high as an astonishing $1,000 per user per month (cited in a presentation at Software 2007).  Quoting in full:

“Here are some numbers. If you spend $ 1 on a software license, over the next decade you will spend between $ 2 and 3 in annual maintenance contracts. You will spend 50c to $ 1 in project team training (vendor classroom training, travel costs, additional licensing of a training tool like RWD that SAP recommends), You will spend between $ 1 and $ 5 with the systems integrator and software vendor consultants (in some cases the services  may even be $ 10+ if you have a complex roll out). You will spend between $ 1 and $ 2 for every major upgrade - likely one every 3 years. So, even without factoring in costs of your own staff or incremental hardware you are looking at $7.50 at the low end and $ 21 (or more) at the high end for every $ 1 dollar of software license costs.”

For Business By Design, SAP’s new, purpose-built SaaS product, the company is claiming pricing in the range of $150 per user month.  But even if that’s a proxy for similar suite products such as Netsuite and Workday, that doesn’t mean the TCO is that low.  For example, just looking at Vinnie’s list above, there are still upfront costs such as project team and user training, business process reengineering, configuration, and integration and data migration from vendor and SI consultants.  Finally, nobody knows how difficult or expensive it will be to maintain the admittedly fewer interfaces to legacy or specialized systems across upgrades than it will be for best of breed SaaS applications.  Interfaces do change, even for SaaS suites, especially as the functionality evolves.  One issue no one disputes is that SaaS makes an enormous difference in the crushing cost of supporting older versions on an extended compatibility matrix of possible customer infrastructure configurations.  In client-server systems, that could chew up as much as 80% of the vendor's R&D budget and untold amounts of customer operations expense.  With SaaS, that all goes into innovation that gets delivered to the customer as soon as it's ready.

Hard evidence of the customer TCO advantage for SaaS suites in the enterprise won’t emerge for a period of at least another 5-7 years we would guess.  Even if the products were mature today, which is far from the truth for reasons we explain below, it would still take considerable time before they could fully penetrate IT operations at some major companies.  Only then, when they can measurably affect the IT cost structure of these customers over time compared to their peers running on different platforms, will all CIOs feel the pressure to meet the new, lower cost benchmarks.

Why SaaS suites need time to develop their go to market approach

Others have written about how for SMBs suite vendors can’t use the more viral adoption model aided by inside sales typical of departmental approaches.  While suites need to reach C-level decision makers, Netsuite, for example, claims that in order to scale more rapidly it needs VAR/SI’s who can make a business out of a first year subscription payment of $25K (vs $75K for Great Plains) and business process reengineering, training, and data migration services of $25-50K (vs $75-225K for Great Plains).  Relative to premise-based software, the service mix has to be lighter.  The only way they believe that model can work is if the VAR/SI’s can create reusable, vertical extensions that Netsuite hosts for which partners can charge another $25K in high margin subscriptions in order to subsidize a $50-$100K implementation project with minimal margins.

The challenge of selling direct with a richer services mix without viral adoption and while trying to build a channel has created entirely different economics for Netsuite vs. SalesForce.com.  Both are roughly 10 years old but Netsuite's estimated revenues for 2008 are $160m vs SalesForce.com's $1080m.  Netsuite lives on -0.05% operating margins compared to SalesForce.com's 15%.  Both stagger under sales & marketing expenses that approximate 50% of revenue because of the slower reported revenue growth coming from subscriptions.  But Netsuite’s gross margin is roughly 10% lower than SalesForce.com's, coming in at 71%, because of the richer service mix currently required to deploy suites.

Two other potential partner models include resellers who can reach a higher volume of customers in order to live on the lower margin commissions, or IT outsourcers who take over a customer’s entire IT shop, including SaaS applications, and become a managed service provider (MSP) for the whole infrastructure.

At the enterprise level, despite being able to charge customers for upfront commitments of one or two years given enough of a discount, the recognized revenue still gets pushed out, depressing growth and margins.  Bookings growth isn’t a substitute.  The only way to make this work is to gradually migrate revenues over to the new model over time.  Plus, vendors will have to overcome additional partner resistance.  The rich time & materials billings generated by custom implementations has to give way to a new generation of highly configurable software.

Why SaaS suites need time to mature as products

While this topic is worth a whole blog post on its own, several issues pop out.  First, there is the issue of the trade-offs between multi-tenancy and configurability.  Vendors such as SalesForce.com appear to tackle the problem of configurability and multi-tenant scalability purely in shared tables within one database schema instance.  They appear to have cracked the code on scaling out with commodity boxes rather than scaling up with expensive big iron while keeping the marginal cost of administration down.

It’s still unclear if they are disadvantaged in configurability.

SAP and Oracle, whether by virtue of having to drag along legacy technology or some other reason, appear to offer customers more isolation.  SAP gives each customer their own database instance and Oracle's On Demand product appears to give each customer an entire application instance.  Keeping marginal administration costs low falls to the maturing management layer, the same type of technology that VMware claims will enable applications to meet service level agreements autonomously (e.g. without administrator intervention).  Like SalesForce.com on a different dimension, it's still unclear if they are disadvantaged on the ultimate TCO front by virtue of high internal administration costs.  Giving each customer their own instance, however, has the added benefit of reassuring them that their data won’t be commingled with anyone else’s.

The other issue holding up suite maturity is simply the Mythical Man Month.  Suites take time to reach functional maturity no matter how many developers are working on them.  There have been reports in the blogosphere of complaints in the Netsuite customer base about product maturity 10 years after they started.  And SAP co-founder Hasso Plattner told a questioner at a Churchill Club event this spring that he would have to wait another 5 years for an enterprise-ready SaaS product from SAP, despite having started 4-5 years ago.

SaaS relevant for a class of processes today, but not all

For suites to reach both SMB’s and the enterprise, there needs to be both business model and product innovation.  Given the customer popularity and VC attention, many startups appear to believe that subscription pricing and hosting means they are SaaS vendors.  That just qualifies them as warmed over application service providers (ASPs).  This transition seems to be shaping up as a classic Innovator’s Dilemma.  Oracle says SaaS isn’t attractive because it’s a lower margin business that requires a new product and a new go to market model.  SAP seems to have reached the same conclusion but is plowing ahead gradually on both fronts.  While SaaS clearly appears to be working for departmental applications, the jury will be out on enterprise-class suites for several years at least.  Everyone agrees you can’t take an existing enterprise-class client-server suite and squeeze enough complexity out of it to reach SMB’s.  Nor, on the go to market side, can enterprise vendors immediately substitute a subscription pricing model for their traditional perpetual license one.  Netsuite seems unlikely to reach the enterprise by virtue of its design center and channel.  Meanwhile, Workday appears to be following the same steady strategy into the enterprise as its predecessor, PeopleSoft, without taking on SAP and Oracle in manufacturing this time.  In the end we believe the Innovator’s Solution in the enterprise is going to take some years more both on the product and go to market side.

I did spend a while trying to figure out why the Schema wasn't appearing but I guess it's just another quirk of Oracle I'll have to get used to.

I wrote a script that can be run to quickly free up space. It connects to the ASM instance and removes the logfiles.

It sets the SID and ORACLE_HOME to that of the RAC instance then runs RMAN to perform a crosscheck and delete of the archivelogs.

Not particularly complex but efficient.

export ORACLE_SID=+ASM1
export ORACLE_HOME=/u00/app/asm/product/11.1.0/db_1

asmcmd -p << EOF

ls FRA/RACCLUSTER/ARCHIVELOG/*2008*/*
rm -rf FRA/RACCLUSTER/ARCHIVELOG/*
ls FRA/RACCLUSTER/ARCHIVELOG/*
EOF

export ORACLE_SID=RACSID
export ORACLE_HOME=/u00/app/oracle/product/11.1.0/db_2

rman target / catalog username/password@catdb << EOF1

CHANGE ARCHIVELOG ALL VALIDATE;
DELETE NOPROMPT EXPIRED ARCHIVELOG ALL;
EOF1

]]> http://jamesrossiter.wordpress.com/?p=24 Thu, 17 Jul 2008 10:14:11 +0000 rossinio http://jamesrossiter.wordpress.com/?p=24 If you set the TMP and TMPDIR environment variables in /etc/profile to a directory that does not exist Solaris will NOT load the desktop environment and you will not be able to log in. There is no error or warning, you just get booted back to the login screen.

Spotted this in an Oracle install when I was testing all the environment variables before actually creating the directories. The oracle directory I set in ORACLE_BASE does not need to actually exist when you log in as it is not used by Solaris itself but the TMP/TMPDIR directory does.

At some point I will write a step by step guide to installing Oracle 10g on Solaris x86 inside VirtualBox, at least then I will have something to follow myself without making mistakes like this!