On Formating Treasures of Execution Plan Interpretation

I usually use SQL*Plus’s AUTOTRACE option for a quick understanding of some query’s execution steps. But with DBMS_XPLAN’s formatting options someone can find much more detailed information which may shorten a tuning workshop.

set serveroutput off        
set linesize 2500
set autotrace traceonly explain

select /*+ gather_plan_statistics */ 
       first_name, salary, department_name
  from employees e, departments d
 where e.department_id = d.department_id
   and d.department_name like 'A%' ;

Execution Plan
Plan hash value: 1021246405

| Id  | Operation                    | Name              | Rows  | Bytes | Cost (%CPU)| Time     |
|   0 | SELECT STATEMENT             |                   |    10 |   300 |     4   (0)| 00:00:01 |
|   1 |  NESTED LOOPS                |                   |       |       |            |          |
|   2 |   NESTED LOOPS               |                   |    10 |   300 |     4   (0)| 00:00:01 |
|*  3 |    TABLE ACCESS FULL         | DEPARTMENTS       |     1 |    16 |     3   (0)| 00:00:01 |
|*  4 |    INDEX RANGE SCAN          | EMP_DEPARTMENT_IX |    10 |       |     0   (0)| 00:00:01 |
|   5 |   TABLE ACCESS BY INDEX ROWID| EMPLOYEES         |    10 |   140 |     1   (0)| 00:00:01 |

Predicate Information (identified by operation id):

   3 - filter("D"."DEPARTMENT_NAME" LIKE 'A%')
   4 - access("E"."DEPARTMENT_ID"="D"."DEPARTMENT_ID")

set autot off
select /*+ gather_plan_statistics */ 
       first_name, salary, department_name
  from employees e, departments d
 where e.department_id = d.department_id
   and d.department_name like 'A%' ;

select * from table(dbms_xplan.display_cursor(null, null, 'TYPICAL IOSTATS LAST')) ; 

SQL_ID  134hnsg0n3tr6, child number 0
select /*+ gather_plan_statistics */        first_name, salary,
department_name   from employees e, departments d  where
e.department_id = d.department_id    and d.department_name like 'A%'

Plan hash value: 1021246405

| Id  | Operation                    | Name              | Starts | E-Rows |E-Bytes| Cost (%CPU)| E-Time   | A-Rows |   A-Time   | Buffers | Reads  |

|   1 |  NESTED LOOPS                |                   |      1 |        |       |            |          |      3 |00:00:00.08 |      13 |      8 |
|   2 |   NESTED LOOPS               |                   |      1 |     10 |   300 |     4   (0)| 00:00:01 |      3 |00:00:00.08 |      11 |      7 |
|*  3 |    TABLE ACCESS FULL         | DEPARTMENTS       |      1 |      1 |    16 |     3   (0)| 00:00:01 |      2 |00:00:00.06 |       8 |      6 |
|*  4 |    INDEX RANGE SCAN          | EMP_DEPARTMENT_IX |      2 |     10 |       |     0   (0)|          |      3 |00:00:00.02 |       3 |      1 |
|   5 |   TABLE ACCESS BY INDEX ROWID| EMPLOYEES         |      3 |     10 |   140 |     1   (0)| 00:00:01 |      3 |00:00:00.01 |       2 |      1 |

Predicate Information (identified by operation id):

   3 - filter("D"."DEPARTMENT_NAME" LIKE 'A%')

   4 - access("E"."DEPARTMENT_ID"="D"."DEPARTMENT_ID")

24 rows selected.

In above example I used GATHER_PLAN_STATISTICS hint to force row source execution statistics(check V$STATISTICS_LEVEL) generation, the alternative is to set STATISTICS_LEVEL parameter to ALL at session level, as a result now you have very important additional information compared to the standard AUTOTRACE output, like;
– “A-Rows” – the number of rows actually produced by the corresponding row source(remember the ‘tuning by cardinality‘ paper by Mr.Wolfgang Brietling)
– “Buffers” – the number of consistent reads done by the row source
– “Starts” – indicating how many times the corresponding operation was repeated

Unfortunately in my opinion information on these options are limited in the documentation here, so for more detailed explanation reading Mr.Jonathan Lewis’s post here may be good starting point.


OOW08 Presentation Downloads

I checked and couldn’t see some of my favorite sessions’ materials available for download yet, but still you may want to start consuming incrementally already uploaded presentations. :)

For starters let me share my three favorite tracks:

Data Warehouse Track – http://tinyurl.com/4puv4b

Application Development Track – http://tinyurl.com/3mwmyp

Performance and Scalability Track – http://tinyurl.com/474ul7

ps: Only OOW08 attendants are allowed to download unfortunately, so after you logon you can use Username: cboracle & Password: oraclec6

11g’s Metadata Information for SQL Built-In Operators and Functions

11g introduced two new views: V$SQLFN_METADATA and V$SQLFN_ARG_METADATA, which provide metadata for all Oracle SQL built-in operations and functions.

SQL> set linesize 2500
SQL> SELECT * FROM v$sqlfn_metadata where name = 'REGEXP_SUBSTR' ;

   FUNC_ID NAME                              MINARGS    MAXARGS DATATYPE VERSION      ANA AGG DISP_TYPE     USAGE                          DESCR
---------- ------------------------------ ---------- ---------- -------- ------------ --- --- ------------- ------------------------------ ------------------------------
       526 REGEXP_SUBSTR                           2          5 STRING   V10 Oracle   NO  NO  NORMAL

SQL> SELECT * FROM v$sqlfn_arg_metadata WHERE func_id = 526 ;

---------- ---------- -------- ------------------------------
       526          1 STRING
       526          2 STRING
       526          3 NUMERIC
       526          4 NUMERIC
       526          5 STRING

This new feature may help especially to third party tools to maintain the function usage metadata redundantly in the application layer. For more informations you may check the documentation and this oracle-developer.net article.

11g Enhancement for ALTER TABLE .. ADD COLUMN Functionality

Before Oracle 11g adding new columns with DEFAULT values and NOT NULL constraint required both an exclusive lock on the table and the default value to be stored in all existing records.

Now in Oracle 11g the database can optimize the resource usage and storage requirements for this operation, the default value is stored in the data dictionary instead of updating the table column as a result especially for large tables the execution time is reduced and space is saved.

In addition, the following ADD COLUMN operations can now run concurrently with DML operations:
* Add a NOT NULL column with a default value
* Add a nullable column without a default value
* Add a virtual column

release 1002000300 –

drop table tst_source purge ;
create table tst_source nologging as
select rownum id, text
  from dba_source;

set timing on


Elapsed: 00:00:30.43

set timing off

exec dbms_stats.gather_table_stats(user, 'tst_source');

select count(*) from tst_source where name is NULL ;


release 1101000600 –

drop table tst_source purge ;
create table tst_source nologging as
select rownum id, text
  from dba_source;

set timing on


Elapsed: 00:00:00.10

set timing off

exec dbms_stats.gather_table_stats(user, 'tst_source');

select count(*) from tst_source where name is NULL ;


On SAGE and Oracle’s new 11g SQL Tuning Workshop Education Content

Most probably you think SAGE (this was its project name before Larry’s announcement, much more compact naming so I prefer it :) –Oracle Exadata Storage Server and the HP Oracle Database Machine– will not make a change in your daily life and you maybe unsatisfied with this announcement but I promise to give you something cool whether you are a massive OLAP or an XE developer with this post. :)

But first let me share with you one of my friend’s, who is a warehouse developer, first response to SAGE: “Oh I knew Oracle should work much more faster for years, the fault was not ours you see!” .:) So does these new features promise to reduce the importance of the warehouse development best practices? I really don’t think so..

So on my 22 hours way back home I had a chance to look at this new Oracle University education and I strongly think that this is the best content you may get from a tuning education. First, with this post, let me briefly explain the table of contents and then I think I will be doing some series of blog posts parallel to this content.

If you had 10g’s SQL Tuning Workshop here is not only new features but new chapters in the 11g’s content now, like a whole chapter dedicated to Star Transformation and a very long chapter dedicated to the optimizer operations like access path options of the Oracle database. I tried to motivate developers around me for not doing any SQL Tuning, just do tuning as they are developing and I am very happy to see this chapter as it is a part I always dreamed of. :)

Let me mention another interesting observation of mine, if you attended Jonathan Lewis’s optimizer seminars I think you will also think that the content and even some pictures are very similar in this education to that content, to my understanding this is a kind of a kind approval for people like Mr.Lewis who are trying to improve the communities understanding of the complex technologies like optimizer. No pain no gain right, so I do not think this education now will be easily consumed by starters as it should be.

By the way, as we own one of the largest warehouses in the world and there is a very important possibility that we may be testing and using Exadata and so I may be blogging something more than “reading billions of rows in seconds” at the end of the day ETL(I mean lots of writes), sort operations parallel scans and hash joins are in total what will make the difference since lots of reports(I mean reads) are already tuned to be fast enough by logical design and access path options like partitioning, bitmap indexes and materialized views in today’s optimized warehouses.

Highlights of my last two days at OOW08 and a wrap-up

I am back now back home to İstanbul after another ~22 hours travel, hundreds of emails waiting to be read and replied but still I am happy with these alternative costs of this year’s open world. :)

First Wednesday’s sessions;

Sess.1 – Oracle Advanced Compression: Throw away half of your disks and run your database faster by Sushil Kumar and Vineet Marwah

– Compression is not only beneficial for storage but maybe much more important benefits you gain from the network, memory and backup&recovery resource efficiency.
– OLTP Table Compression of 11gR1 Best Practices; first choose your 10-20 largest tables to compress, better results usually with larger block sizes.
– Compression Advisor is now available at OTN for download DBMS_TABCOMP.GET_RATIO, only work for 11g currently but 9i&10g is on the way.
– Since compression is highly dependent on your dataset testing with this package is highly advised.

Sess.2 – Get the best out of Partitioning: Things you always wanted you know

Case scenarios of 11g’s new partitioning options were demosntrated on this session.

Sess.3 – Change Change Change by Tom Kyte

11g change management new features were discussed on this session.

Fun parts of the day were, during the lunch break I went to the head quarters of Oracle with Route 12 shuttle, had chance to take lots of pictures. And at night at the appreciation event UB40 concert was really cool. :)

Thursday morning I attended my last session at OOW but before the session I had the privilege to meet Mr.Hans Forbich at the OTN Lounge and after the session I had a quick chat with Fuad Arshad just before I left for my flight.

New DB Accelerator Query Processing Revolutionized by Juan Loalza

Below are some keywords from this session, but I am sure we will be talking about this revolution for sometime. :)

– More and bigger pipes for data and CBO like intelligence inside hardware.
– Sold by Oracle, HW supported by HP
– Infiniband protocol, OEL 5.1 Linux
– does not support non-db storage, pure Oracle :) HPDL180G5, fully Oracle optimized, certified and supported configuration
– three tiers: Oracle DB – ASM – Exadata
– iDB ~ iSCSI but additionaly have extensive DB intelligence
– storage grid monitoring and management on grid control
– faster tablespace creation
– smartscans of exadata calls for queries, backups, join filtering and smart resource management
– compute intensive functionality still in the DB versus data intensive processing on exadata(not another DB node in short)

As a wrap-up it was a completely different open world experience for me this year, very well organized and technical content was much more in a balance with the typical marketing stuff, now time to wait for the sessions’ presentations download. :)

Highlights of my third day at OOW2008

Hopefully I started to adopt to SF and now feeling a lot better, Tuesday was my best day until now and mostly focused with again the datawarehouse sessions and meeting some old friends from Oracle blogging community.

I have two quick hints to share: I learned that there were %20-25 discounts with Oracle books at the bookstore, West Level 2, I immediately bought several books and gifts to my Oracle database geek friends at İstanbul.

Also I learned that there are periodic shuttles to the airport and the Oracle head quarters from Moscone Center, I plan to use try of them, who may miss the chance to have some pictures for facebook near the boss’s ship parked at the lake in front of the cylindrical buildings of Oracle head quarters? :)

Sess.1 – Insider 11g Optimizer: Removing the Mystery by Maria Colgan and Mohammed Zait

This was another very well organized session from Maria related to the new optimizer features, most of them designed to end the pains we face for years, and yes they are strong and very promising this time. We have used 11gR1 database and 11g OWB at our ODS(Operational Data Store) implementation and I had opportunity to use some 11g NFs, yes there are several nasty bugs but if you manage your support issues with close relationship I strongly believe 11g is the best platform for both a warehouse developer and a dba to be.

One may think why Oracle is so proud to announce these stuff after causing all those years of pain at its customers, but lets remember that optimizer development is something to admire, it is a real software engineering process, one piece of smart box to rule any kind of Oracle database customer need, that’s why I believe it is very acceptable for CBO to be still evolving.

Sess.2 – The Best Way Keynote by Tom Kyte

Tom is one of my all time heroes, it is always a pleasure to listen to him, but I guess I am reading his books, asktom threads and Oramag articles so much that this session was a little boring for me. :) After the session I met Eddie Awad, it was a real pleasure to meet Eddie for me, I admire the things he has done to support the Oracle community, I believe if we manage to become a strong community he is one of the major actors in this evolution.

Sess.3 – Advanced performance diagnostics: What GUIs doesn’t tell you by Kurl Engeiler and Jon Waldron

There were several EM, AWR, ASH comparison type of information and practical advises like trend analysis of AWR and ASH data in this session, I liked it. patch now has the EM support for Real Time SQL Monitoring feature of 11g. spawrrac.sql can be used to have a global text based AWR report for RAC on 11g.

Sess.4 – 11g Stories from a Datawarehouse Implementation by Dr.Hulger Friedrich

This was my best session for the day, Mr.Friedrich shared an example OWB 11g framework on Expert Editor to develop a parametric mapping developer, ETL process is usually can be managed to fit in this type of a smart framework so a lot of development efforts will be dropped, it was a very cool best practice demonstration, my team back at Turkcell will love this one I am pretty sure. :)

Sess.5 – Hand-on Lab: SQL Developer Data Modeling Feature

I had the pleasure to meet Sue Harper and ask questions to her, like XE and APEX I love and advocate SQL Developer within my organization, I love idea of Oracle investing in these *no-additional-cost* database technologies, but unfortunately data modeling feature of SQL Developer seems to come out with an extra license. This feature is coming out enough strong depending to my hands-on lab, it is Oracle database aware of course in physical design but also it is OLAP aware, nice to have it around and hear Oracle will be investing in it in the future more. As far I understand other than our Data Modeling needs Jdeveloper will be the platform for us.

After the session I stopped at the near salon for a quick chat with Patrick Wolf after his session and spent some time at the UNCONFERENCE session of Dan Norris: So, you want to be an Oracle ACE. I know some of my all time idols left ACE program, the main reason behind I guess is first this title was assumed to have close relationship with some kind of excellent Oracle knowledge, but the being ACE idea is not directly related to this as far as my understanding, it is mostly related to what you give, share with the community and these are usually not related to being so deep technical but web 2.0 type of actions, still very very beneficial actions of course.

Also I guess there is no path to walk to be an ACE, it is some kind of an outcome of your choices, which to spend a lot of your time taking some actions for the community, like blogging or answering to threads at OTN forums. And finally someday you will learn that you gained this award. :)

Let me share my experience with you with this opportunity, it was early 2006 I guess, I broke up with my girlfriend after a long time relationship and at those days I was not able to sleep at all, and all I loved to do was blogging and foruming until the daytime, this depression took nearly 1,5 year, I gained ~20 kg and get addicted to blogging and foruming. And someday I recieved an email, funny isn’t it. :) Also I guess not being an ACE wouldn’t change anything in my life for today, I would continue what I did for years mostly because I love doing this. Dan’s session was important to point out these kind of questions.

One quick comment on the UNCONFERENCE sessions, I think we(bloggers) have to market these sessions more, Dan’s session had a very limited attendance similar to my APEX session unfortunately, where as I am pretty sure a massive amount of people could be interested to have these two sessions.

Sess.5 – my ACE hours at OTN Lounge

I put my picture on the map at the wall immediately when I saw my country’s space was empty, it is fun if you still didn’t try it yet, and yes I admit I missed İstanbul a lot. :)

I had great time standing and chatting to Rob, Nicholas, Chen and Andreas, feels a little weird to meet face to face to people you feel like you are so close after all those time of blogging and at OTN forums, thank to this Openworld opportunity. :)

Sess.6 – Practical DWH Experiences with 11g by Travidis

This was yet another helpful, practical session on OWB 11g and some 11g database DWH new features.

At night I couldn’t attend to the annual ACE dinner because I promised to attend to the Terabyte Club dinner, I was very unhappy to loose the chance to meet some of my heroes like Mr.Hans Forbich, Mr.Dan Morgan and Mr.Arup Nanda meanwhile it was a great experience to dine with Oracle’s datawarehouse product managers and Amazon warehouse architects, as far as I listened Amazon really rocks in terms of System and Software Engineering, they have the attribute I always dreamed of, I admired to their DWH architecture, vendor management strategy and the way they handle problems.

Today is the long waited *X* day, maybe better to call it the *XXXX* day. :)P XXXX is really big, especially if you are in the OLAP world, so do not miss Larry’s keynote today, most probably next openworld we the Terabyte Club members will be presenting how we benefited from this rocking feature. This is all I will be saying for now to motivate you. :)

Tomorrow I will be leaving immediately to catch my flight after the first “Data Accelerator” session of the last day, so I hope to make the best out of my last day here, hopefully see you tomorrow with another busy day summary I guess. :)


Highlights of my second day at OOW2008

Monday was a very busy day for me, mostly focused with the datawarehouse sessions.

Sess.1 – UNCONFERENCE – APEX Test Drive by me :)

I guess because of it was a morning session the attandance was very limited to my session. Also I had hard time with the sun in the room, it was a very beautiful day by the way outside on Monday.

Here is my presentation, I hope you also have fun like I did during preparing it. :)

Sess.2 – Growing a DWH to 50 TBs and beyond by Hüsnü Şensoy

After my session I listened to the Keynotes, met with some Terabyte Club members to chat and moved to my collegue’s session. During the session I met Julian Dyke and chat, I was suprised to see an internals experts around here. :)

Hüsnü shared our experiences with Oracle products in our BIS Re-Engineering project at Turkcell, it was a very helpful, practical content for the attandents I guess. He will be sharing the content in his blog so if you are intrested you may add his RSS to your reader.

Sess.3 & 4 – Run a 100 TB DWH on RAC and Linux by Amazon & DWH at 300 TB on RAC by LGR Telecom(At&t-Vodafone..)

Both sessions were informative but I was very suprised to learn that LGR was not using ASM nor RMAN. Also these kinds of sessions must be done by doers not decision makers in my opinion, I want to hear the DBAs experiences not his managers.

Sess.5 – Best practices for deploying a DWH on 11g by Maria Calgan(Principal Product Manage)

This presentation was impressive, all DWH Oracle best practices within 50 minutes, I felt very lucky to be inside. Session was divided into four parts: Hardware, Logical Model, Physical Model, System Management.

I strongly advise you to download both her presentation after the conference and the OTN paper related to this session, even you are not on 11g nor a DBA of a DWH. Lessions learned for me: don’t waste your time with customers’ presentation but try to find some Oracle Product Manager session. Anyway I learned that they are at the Demogrounds and I will be visiting them.

Sess.6 – Partitioning with 11g by Bert Scalzo(Quest)

I was feeling so tired that I couldn’t finish this session and when I went back to the hotel I slept like 12 hours and missed OTN night, still feeling not so good this morning and planned to have another very busy day for Tuesday sessions, so see you tomorrow if I still can survive to write another highlight blog, and today I will at OTN Lounge(Moscone West Level 3) for my ACE Hour session if you are intrested.


Highlights of my first day at OOW2008

It took ~22 hours for me to get to Cathedral Hotel on OOW’s Route 2 from my home, everything is cool until now. OOW 2008 is well organized and the city is like Oracle’s home, Oracle is everywhere here and below are the highlights from my first day at OOW 2008.

Sess.1 – Getting started with Oracle Fusion Middleware by George Trujillo

During this presentation I sensed George was trying to underline the next big thing, the change we will soon face as the database developers and DBAs. Call Forms and Reports developers Developer 1.0, then Fusion Middleware developers are the Developer 2.0 era, here key components are XML, SOA, BPEL, Web Services and J2EE. XML is like English, it is the common language all applications must speak in order to integrate with each other in this architecture. Oracle’s Weblogic purchase was an important step, within Oracle Weblogic Application Server SOA Suite, Identity Management, Business Intelligence and User Interface, Portal, ADF, JSF will be the four major areas of interest. All managed with a single point of view, by Grid Control, similar to the RAC environment at the data tier.

To my experience these technologies’ experts all stand very far from the database so if Fusion Middleware and these guys are the future I suspect there will be big threats on building successful database applications, but of course on the other hand for experts who can additionally manage to learn these technologies there will be important opportunities to fix these problems.

Sess.2 – Extending Oracle Application Express Framework with Web 2.0

This was a hands on lab based on OTN OBE’s for Javascript and Ajax usage within Apex 3.1.

Sess.3 – All About Encryption by Tom Kyte

This was another very typical Kyte presentation, with demos on performance and storage effects of the encryption options. Presentation’s focus was three features inside the database;

a. DBMS_CRYPTO: no additional cost, labor intensive, not transparent, not performant when compared to the other two options.

b. Column Level Encryption: Advanced Security Option needed, semi transparent, can not use index with LIKE ‘ABC%’ type of queries, storage cost is higher after encryption, much more performant compared to DBMS_CRYPTO.

b. Tablespace Level Encryption: Advanced Security Option needed, completely transparent, performant on both read and writes and also disk usage doesn’t increase.

With DBMS_OBFUSCATION_TOOLKIT instead of using VARCHAR APIS it is better to convert to RAW first and always return BINARY formatted data. After 11g Data Pump can export encrypted and preserve the protection, leave old EXP it will be depreciated soon, but IMP will stay. Undo, Redo produced and Data cached all are encrypted also, but a last minute question was related to Temp(Sort) data and Kyte promised to blog on this topic. Becareful about the legacy information, old backups for example, encrypting today will not protect the data sitting there.

Sess.4 – Doing SQL from PL/SQL: Best and Worst Practices by Bryn Llewellyn

There is a new OTN paper on this session, Bryn mentioned some very interesting PL/SQL Best Practices during this session. This was my best session for the day.

Sess.5 – 11g New Features Exam Cram

I planned to attend Kyte’s Efficient Schema Design session but changed my mind to move to the Exam Cram, within 8 hours all 5 days 11gR1 NF education topics were discussed, it was cool to attend the last 2,5 hours, but I couldn’t have the booklet prepared for the cram, hope to get one tomorrow.

I had the chance to chat with three Oracle bloggers until now; Tim Hall at the plane, Carl Backstrom and Steven Feuerstein during the sessions today, but now it is time for the bloggers meetup of OOW 2008 but everything is turning around my head tonight after the first 1,5 days, I hope I can make it. :)

ps: I tried to twit my sessions as they happened today and will try to continue this until Thursday, so if you want to follow here is the link; http://twitter.com/TongucY

OOW 2008 journey starts tonight

It has been nearly two months now that I have been preparing for the OOW 2008;


This year I tried to focus on the warehousing sessions and the opportunity to meet with the friends from the community. It came up to be a very busy schedule for 5 days, I guess I will not have any extra time to shop or gather around SF. At 26th I will be back in Istanbul so I will be writing a single wrap-up about my experiences next weekend I guess. Here is the probable Table of Contents. :)

– At Sunday 21, I will visit OCP Lounge for the 11g Exam Cram, I am onto an experiment, I will be having a very limited time to study the NFs of 11gR1 at the plane, lets wait and see the result. During blogger meet-up at night I hope not to experience a intensive jet-lag and enjoy this time with my favorite Oracle bloggers as long as possible.

– At Monday 22, I will do an Unconference session; Oracle Application Express Test Drive for DBAs and PL/SQL Developers, I will share this presentation with my promised wrap-up. At night this time OTN Night will be the place to have fun.

– At Tuesday 23, I will be at the OTN Lounge for the Oracle ACE Office Hours, I hope to meet and chat with some of my friends again with this opportunity. This night is very complicated, there will be three dinners I want to attend; Terabyte Club appreciation dinner, annual Oracle ACE dinner and Turkish Oracle Users dinner. I am sorry that I will be able to make only one of them.

– At Wednesday 24, I will be focusing on 11g warehousing, meet some TB Club member DBAs to exchange some experiences. And at night I will be having fun at the Appreciation Event.

Thursday morning I will be packing up and my plane will be leaving SF at lunch time. So here are some quick links you may want to check if you are also attending to OOW 2008, hope to meet you there. :)

OOW 2008 Portal

OOW 2008 Blog

OOW 2008 A quick review of the sessions

Oracle OpenWorld and Oracle Develop 2008 Events Portal

OOW 2008 Unconference

OOW 2008 11g Exam Cram Session

OOW 2008 ACE hours at OTN Lounge