MySQL Lists are EOL. Please join:

List:Cluster« Previous MessageNext Message »
From:Mikael Ronström Date:July 12 2005 8:56am
Subject:Re: Large Table maintainance
View as plain text  
Hi Adam,
MySQL Cluster handles deletions well but is a main memory solution 
So it is quite ok to use it for an amount of data and then move it to 
table when you want it to go off-line.

In 5.1 that will arrive in a short time in first alpha releases you 
will have two
1) MySQL Cluster will be able to store non-indexed fields on disk
2) Your MyISAM tables can be partitioned.

By partitioning you can delete a partition without affecting other 
You will also be able to move a partition to a new table and this will 
be close
to immediate.

There is a forum for partitioning that you can check for latest status.

Rgrds Mikael
PS: Most disk-based indexes have problems with deletions, to reorganise 
immediately after a delete is complex and most DBMS's have opted to 
avoid that
complexity. So the index is still ok but not as efficient since the 
packing degree is

2005-07-12 kl. 07.17 skrev Adam Dixon:

> I have a large table containing 12 varchar, 19 int fields. This table
> consitantly grows at around 7000 records per hour. With the current
> MyISAM solution on a single server - if we delete older rows from the
> table, it causes queries to this table (which are done very
> frequently) to slowdown considerably - There is probably a technical
> explaination to this;
> Currently we move the old data in an offline state to another table,
> then we optimize the remaining recent data, which takes along time.
> So basically using DELETE on the table ruins its speed, can anyone
> comment as to if NDBCLUSTER handles the deletion of rows better then
> MyISAM? So that we can move older data to a seperate system.
> I would love to do a 'on the fly' routine to clear up old data, eg
> select and insert into a offline area so that we can maintain the size
> of the cluster.
>> From what I can gather the index's become ineffective if delete is
> used on a myisam table.
> Or if anyone has any suggestions for this comments would be great.
> Adam
> -- 
> MySQL Cluster Mailing List
> For list archives:
> To unsubscribe:    
Mikael Ronstrom, Senior Software Architect

Jumpstart your cluster:

Large Table maintainanceAdam Dixon12 Jul
  • Re: Large Table maintainanceStewart Smith12 Jul
    • Re: Large Table maintainanceAdam Dixon12 Jul
      • Re: Large Table maintainanceMikael Ronström12 Jul
  • Re: Large Table maintainanceMikael Ronström12 Jul
    • RE: Large Table maintainanceJonathan Miller12 Jul