PROVOCATIVE RANT
MySQL doesn't provide native support for transactions but is dependent on
the Berkeley or InnoDB table types.
Does MySQL really provide TRUE row level locking? That's the claim but is it
valid? The InnoDB web site seems to be more accurate by explaining that
locking is a
Background:
I've developed a simplistic Perl program to test database performance with
concurrent session queries. The queries involve inserts, updates, and
deletes in order to test database performance in an OLTP mult-user ACID
compliant scenario. Obviously this is not a real world test but it
-
From: Orr, Steve [EMAIL PROTECTED]
Newsgroups: mailing.database.mysql
Sent: Thursday, September 05, 2002 5:52 PM
Subject: Performance Problems with InnoDB Row Level Locking...
Background:
I've developed a simplistic Perl program to test database performance with
concurrent session queries
: Heikki Tuuri [mailto:[EMAIL PROTECTED]]
Sent: Thursday, September 05, 2002 2:54 PM
To: Orr, Steve
Cc: [EMAIL PROTECTED]
Subject: Re: Performance Problems with InnoDB Row Level Locking...
Steve,
- Original Message -
From: Orr, Steve [EMAIL PROTECTED]
To: 'Heikki Tuuri' [EMAIL PROTECTED
I profiled my PHP app with a debugger
A single session and a few queries does not make a very good test. The
reason for persistent connections is to scale web apps to handle intense
multi-user activity like 1000 queries per second or 100's or even 1000's of
concurrent database connections. Do a
There is at least one reason, which is that persistent connections cause
the server to hold open connection slots even the connection isn't
actively
being used at the moment.
But isn't this really just a failure of the connection pooling mechanism? If
you have a connection pooling program on
Regarding your testing... did you test many concurrent processes or
concurrent multi-user access? If you need concurrent multi-user access for
an OLTP app then you should test it.
To this end I developed a Perl routine which launched concurrent sessions in
the background. I could adjust the
If you've got to make copies to disk then why not just maintain copies on
disk via replication? Stop the slave, backup the slave, restart the slave.
-Original Message-
From: Nicholas Stuart [mailto:[EMAIL PROTECTED]]
Sent: Wednesday, July 31, 2002 8:55 AM
To: [EMAIL PROTECTED]; [EMAIL
It's my understanding that MySQL will only use one index per table on a
given query. For example...
SELECT * FROM HUGE_TABLE
WHERE col1 = val1
AND col2 val2
AND col3 val3 ;
If col1, col2, and col3 are indexed the query can only use one index, right?
Single index access is a problem when you
InnoDB COULD help if there's an issue with table locking and intense
concurrent multi-user access. InnoDB's row level locking should help with
this. Some queries will be faster with InnoDB and some will be slower. For
instance, select count(*) from table_name will do a full table scan with
A slave can only have one master.
You obviously don't work around here. :-)
-Original Message-
From: Jeremy Zawodny [mailto:[EMAIL PROTECTED]]
Subject: Re: Multiple masters to 1 slave??
On Tue, Jul 02, 2002 at 10:54:44AM -0500, Jim Crippen wrote:
In the my.cnf file on the off-site
It's not THAT much like Oracle. ;-)
-Original Message-
From: Silmara Cristina Basso [mailto:[EMAIL PROTECTED]]
Sent: Tuesday, June 18, 2002 9:16 AM
To: [EMAIL PROTECTED]; Egor Egorov
Subject: Re: create a new table in a tablespace...
If i had more than one tablespace, don't i control
The port of Cartesia does ship a lot of product. ;-)
-Original Message-
From: Norris, Joseph [mailto:[EMAIL PROTECTED]]
Sent: Tuesday, June 18, 2002 1:02 PM
To: 'Benjamin Pflugmann'; Norris, Joseph
Cc: Mysql_List (E-mail)
Subject: RE: HELP!!! major situation with mysql on windows 2000
,
Heikki
- Original Message -
From: Orr, Steve [EMAIL PROTECTED]
Newsgroups: mailing.database.mysql
Sent: Wednesday, June 12, 2002 8:20 PM
Subject: MySQL/InnoDB Hot Backups - What's a binlog segment?
I'm confused about the meaning of the help text from ibbackup --help.
Here's the text
:-)
-Original Message-
From: Kiss Dániel [mailto:[EMAIL PROTECTED]]
Sent: Friday, June 07, 2002 12:04 AM
To: Orr, Steve; [EMAIL PROTECTED]
Subject: Re: InnoDB Hot Backups... ALL OR NOTHING ???
First of all, there are many aspects of your problem.
1. The InnoDB uses transaction safe table types
What's the best way to lock down a slave to ensure that only select SQL
queries are executed and that all command line SQL inserts, updates, and
deletes are performed only on the master?
TIA
-
Before posting, please check:
It appears the v4.0.1 mysqlhotcopy perl script has been modified and broken.
When using cp for the backup it's passing a null value for the file(s) to be
copied. Below is a session clipping of an execution with --debug turned on:
#
Can you backup/restore just one MySQL database with InnoDB hot backup? From
what I gather it's an all or nothing proposition. As I understand it,
there's only one tablespace (with any number of data files) and all
database tables of type 'InnoDB' are comingled in the one tablespace.
Therefore, if
I changed my.cnf to add an autoextending datafile but I get errors on
startup.
Here's the my.cnf syntax:
innodb_data_file_path = hat/hatdata1:64M;hat/hatdata2:64M:autoextend
Here's the error output from mysqld:
InnoDB: syntax error in innodb_data_file_path
020523 16:34:24 Can't init databases
Metadata in the database is NOT a bad idea! It's a basic part of the
relational model, Codd's 4th rule. (E. F. Codd is the originator of the
relational model.) While no one has implemented all 12 of Codd's rules, this
is pretty basic/easy.
Here's the 4th rule: The database description is
20 matches
Mail list logo