Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: Database or store to handle 30 Mb/sec and 40,000 inserts/sec

Re: Database or store to handle 30 Mb/sec and 40,000 inserts/sec

From: Serge Rielau <srielau_at_ca.ibm.com>
Date: Mon, 20 Feb 2006 13:55:34 -0500
Message-ID: <45uhlbF8ischU1@individual.net>


Mark Townsend wrote:

> Serge Rielau wrote:
> 

>> Galen Boyer wrote:
>>
>>> On 19 Feb 2006, galen_boyer_at_yahoo.com wrote:
>>>
>>>
>>>> When would you ever want to read uncommitted records?
>>
>>
>> Uncommitted read is just fine for anything statistical.
>> When mining a DSS or ODS system there is no need to get exact data.
>> Whether someone returned a pair of shoes or not is irrelevant for
>> trend analysis.
>> Does Oracle support query sampling? If so, there you go...
> 
> 
> So IBM never recommends UR for DB2 LUW except for situations where users 
> can tolerate wrong results (i.e data sampling ?). 
Just like Oracle never recommends bitmap indexes in OLTP.. So _barring_stupidity_ I'd guess that's correct. I can't warrant for 160,000 IGS employees ;-)

> And presumably you would advise MS users to not use UR as well ? Unless they can tolerate the implication. Just like I wouldn't recommend an Oracle customers snapshot isolation if their transaction must be serializable.

> What is the default isolation level for DB2 LUW ? Cursor stability (aka read committed).
This is also by far the most popular isolation level on DB2.

 > What is the default isolation level for SQL Server 2005 ? I have no clue. But apparently I'm on my way to becoming a SS2005 expert in no time in this group. ;-)

>> I find it highly amusing how posters justify isolation levels based on
>> locking behavior.
>> Isolation is semantics, locking is implementation.
>> There are quite viable solutions for READ COMMITTED isolation level
>> which have the exactly same concurrency behavior as Oracle's
>> implementation of Snapshot Isolation.

> 
> 
> I'm sorry - but who's implementation of RC is the same as Oracle's MVRC ?
And here goes another twist. == Concurrency behavior <> implementation. Not even <> isolation level.

SQL Server 2005, unless i'm gravely mistaken suppors two new twists. One of them is MVRC, the other is last committed without readers blocking writers.
The point being made (at least as I understand it) is that SQL Server's IMPLEMENTATION of MVRC does not scale and is slow for unspecified reasons and that last committed is a bad isolation level BECAUSE of readers blocking writers.

So.. is there any interest in bringing the debate back to a technical discussion?
Obviously I am not trolling for M$. I'm really interested on what folks perceive as the issue.

Cheers
Serge

-- 
Serge Rielau
DB2 Solutions Development
IBM Toronto Lab
Received on Mon Feb 20 2006 - 12:55:34 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US