Thank you for your reply.

I was just wondering that Database Flash Cache could be configured with PCI Cards(Flash Cache) on Exadata Storage Server.

If I doesn’t add addition PCI cards to server, It are no other ways that Database Flash Cache can be configured, right?

there is a statement that makes me confused. If you wouldn’t mind getting asked, I would like to ask you a question.

What I have read in the book is like the following (on Page 108)

“Please do not confuse the Exadata Smart Flash Cache with the Database Flash Cache. Although both use flash technology, the Database Flash Cache is located in the database server node and essentially provides an extension to the standard operations of the SGA. The Exadata Smart Flash Cache, as you will see in the following pages, works in a different manner and is a part of the Exadata Storage Server. In addition, the Database Flash Cache is not supported in the Exadata environment.”

I know the Database Flash Cache is Oracle 11g features, but this book says that the Database Flash Cache is not supported in the Exadata environment.

I would like to know that this means the Database Flash Cache can’t be used In Exadata?, or the Database Flash Cache can’t be used with Exadata Smart Flash Cache feature?

Thank you in advance.

If you have (soft) corruptions in your datablocks (and perhaps in some other cases like just bugs which make oracle think a block is corrupt) then Oracle session who accesses that block will dump some past redo for that buffer – and this means the session has to read from redolog file.

So, the first thing is to check which sessions (foreground or background) are waiting for this event and what are they doing (snapper, ASH, sql trace would be your friends)

I met you several times in HOTSOS conferences (If you could recall someone bothering you with several questions during your lunch :) )

We have online redo log files about 5G with 12 groups multiplexed (2 members in each group).
We changed it to single member groups. Since then we are seeing log file sequential reads as top event.
What causes this event occurs ?

Thanks for your time reading my question.

please take first my apologies for contacting you as owner of the Taner Polder’s Blog. As Oracle database professional you will be familiar with the advantages/disadvantages of Oracle tracefile analysis. So there is no need to waste your time with long explanations.

The company I work for is developing a tool named QueryAdvisor specialized on Oracle tracefile analysis. QueryAdvisor provides more information and allows easier access to tracefile data then TKPROF (alone or in combination with TRCSESS) or even TRCANLZR does.

We are now looking for DBAs and developers assisting as betatesters. We have chosen this approach to be sure that we will cover a huge amount of different Oracle database configurations (different hardware / releases / patch levels).

You will receive a free Administrator License without restrictions or further obligations for your participation in our betatest and/or writing an article. To avoid any misunderstanding – this is not a new approach to sell you something. We are talking about a free license in exchange of some of your time and experience – nothing more nothing less. We hope that this is a win win situation for both of us.

You will find additional information and the possibility to register as betatester at:
http://goo.gl/akWSL

More information how to obtain a free license for writing an article and/or review is available at: http://goo.gl/W2LpF

If you like to do us a special favor – talk to your colleagues about our betatest and share the link.

We will take care that you will not be contacted a second time. Sorry for any inconvenience caused and thanks for your time.

Kind regards
Frank

-dave

1) Unfortunately it’s not possible to convert from SQL_ID to OLD_hash_value as both the 10g SQL_ID and 10g “new” hash_value are computed using MD5 hashing function, but the OLD_hash_value in 9i is computed using the classic Oracle kgghash() function. So you have to match the SQL statement texts (or add a column to your old 9i tables and populate it with new hash value / SQL_ID from the text)

2) If RAT changes the text, then it’s impossible to compare the hash values – I would identify the pattern that RAT always adds (if it is a consistent pattern) and use some regex to remove it and then calculate my own hash values or compare SQL texts directly…

I was attending one of your workshop in Hong Kong last year, very useful. From your website, I have found that you did provide a SQL script to translate a SQL_ID into hash value, this is good in 10g.

My understanding:

in 9i each SQL will have an unique HASH Value
in 10g each SQL will have an unique SQL_ID (your script can translate this into a HASH Value)

The problem is the translated HASH Value in 10g is not the as HASH Value in 9i, that is why in 10g Oracle still keeps a column call “old hash value”, I will like to know the SQL statement to generate such OLD HASH VALUE from the SQL_ID.

Why: We have SQL performance report for application running in 9i, during the period of migration from 9i 10 10g or higher, will like to make comparison between TOP SQL from these 2 environment, therefore if we have the right function to convert SQL_ID to a 9i style hash value, we can easily identify which Top SQL in 9i as well as in 10g.

2) Another question is: when using Real Application Test, The RAT report gives out the TOP SQL with SQL_ID that is not the same SQL_ID as the SQL_ID find in v$sqlarea. Looks like RAT has add some prefix to the SQL and result of that we cannot identify the same Top SQL from AWR report against the RAT report.

Thank you very much

David Tin