<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/">
    <channel>
        <title>The InnoDB Forums - Suggestion Box</title>
        <description>Make suggestions and request new features for any product in the InnoDB family.</description>
        <link>http://forums.innodb.com/list.php?7</link>
        <lastBuildDate>Tue, 07 Sep 2010 14:51:26 +0300</lastBuildDate>
        <generator>Phorum 5.2.13</generator>
        <item>
            <guid>http://forums.innodb.com/read.php?7,978,978#msg-978</guid>
            <title>innodb hot backup  --unapply-log, detach and attach .ibd (innodb_file_per_table) (no replies)</title>
            <link>http://forums.innodb.com/read.php?7,978,978#msg-978</link>
            <description><![CDATA[ Hi,<br />
<br />
There is two things that can be really wonderfull about backup: <br />
<br />
** innodb hot backup<br />
<br />
t1 : start ibbackup<br />
t2 : end ibbackup  ( ibdata and other ibd files are generated, but no ib_logfile)<br />
t3 : start ibbackup --apply-log<br />
t4 : end ibbackup --apply-log (ibdata and others ibd are consistant, and ib_logfile are generated)<br />
<br />
so at t4, the databases backup are identical as the originals at the t3 time.<br />
<br />
Instead of the --apply-log at t3, a command like --unapply-log would may rollback modifications to make the databases backup identical to the originals at time t1.<br />
<br />
** .ibd files, innodb_file_per_table mode<br />
<br />
it would be great if we can detach an ibd file of a table dans attach it on any other,<br />
it would be easy to backup and copy tables.<br />
<br />
especially on a SAN storage with SAN Snapshot<br />
<br />
- the procedure to make a copy of a table would be : <br />
<br />
# CREATE TABLE mytable_save LIKE mytable;<br />
# ALTER  TABLE mytable_save DISCARD TABLESPACE;         -- dettach and drop ibd files<br />
# ALTER  TABLE mytable      DETTACH TABLESPACE;         -- dettach and NOT drop ibd files<br />
# cd /data/database/ ; cp mytable.ibd mytable_save.ibd  -- copie the tablespace <br />
# ALTER  TABLE mytable_save IMPORT  TABLESPACE;         -- attach<br />
<br />
- the procedure to make a backup of the full database stored on a SAN : <br />
<br />
# flush tables with read lock<br />
# snapshot<br />
# unlock tables<br />
<br />
So we could easily mount the snapshot, copy the ibd files and then restore a table.<br />
<br />
I think these two fonctionnality can make MySQL and innodb really powerfull.<br />
<br />
<br />
Victor <br />
DBA MySQL]]></description>
            <dc:creator>vtaing</dc:creator>
            <category>Suggestion Box</category>
            <pubDate>Wed, 06 Jan 2010 19:55:41 +0200</pubDate>
        </item>
        <item>
            <guid>http://forums.innodb.com/read.php?7,713,713#msg-713</guid>
            <title>Confusing/incorrect error message (2 replies)</title>
            <link>http://forums.innodb.com/read.php?7,713,713#msg-713</link>
            <description><![CDATA[ I mistakenly typed --apply-logs instead of --apply-log and the error message I received indicated that the license was for a server with a different hostname and suggested that a license renewal would help.  It would be less confusing if that particular error message simply said &quot;Unknown option: --apply-logs&quot;<br />
<br />
This would probably be helpful for all options but it is particularly relevant for the --apply-log option because it is likely to be run on a server with a different hostname.]]></description>
            <dc:creator>Ladadadada</dc:creator>
            <category>Suggestion Box</category>
            <pubDate>Tue, 24 Aug 2010 18:15:06 +0300</pubDate>
        </item>
        <item>
            <guid>http://forums.innodb.com/read.php?7,591,591#msg-591</guid>
            <title>asynchronous IO (1 reply)</title>
            <link>http://forums.innodb.com/read.php?7,591,591#msg-591</link>
            <description><![CDATA[ I know there was talk years ago about fully implementing asynchronous IO on linux with InnoDB (in 5.1 I believe). Was that work completed?<br />
<br />
Thanks,<br />
<br />
-Ryan]]></description>
            <dc:creator>rhuddleston</dc:creator>
            <category>Suggestion Box</category>
            <pubDate>Thu, 14 May 2009 08:23:40 +0300</pubDate>
        </item>
        <item>
            <guid>http://forums.innodb.com/read.php?7,375,375#msg-375</guid>
            <title>Encryption? (4 replies)</title>
            <link>http://forums.innodb.com/read.php?7,375,375#msg-375</link>
            <description><![CDATA[ I'm developing an application for a business which will use MySQL for data storage, however the client requires that the data be protected from prying eyes should the hardware be compromised. The first idea to come to mind was column encryption - the database password could double as an encryption key and the encrypt() and decrypt() routines could be incorporated into insert, update and select statements - however, column encryption breaks several of MySQL's features.<br />
<br />
I stumbled upon the suggestion of &quot;table-wide encryption&quot;, which, if I'm not mistaken, involves encrypting the entire table at storage time, and decrypting it when it's loaded for use. Which brings me here - could I get some input from someone here on this matter? And would I be wasting my time opening up the source code when there's already a suitable solution available in MySQL/InnoDB which I've overlooked?<br />
I'm also weighing up, should it come to modifying source code, whether to work on modifying InnoDB, or learn enough of MySQL's storage engine API to build my own (inevitably primitive, but hopefully suffficient) storage engine to suit my needs. Neither really sounds pleasant or approachable, but which way should I go?<br />
<br />
A few things to note: the suggestion of drive encryption at the OS level has been made, and I don't like it; and whilst performance would be nice, this project can eschew it if necessary.]]></description>
            <dc:creator>Fenrisulvur</dc:creator>
            <category>Suggestion Box</category>
            <pubDate>Tue, 30 Jun 2009 19:29:18 +0300</pubDate>
        </item>
        <item>
            <guid>http://forums.innodb.com/read.php?7,333,333#msg-333</guid>
            <title>SHOW TABLESPACE :  Reports on space usage and free space available in innodb tablespaces (1 reply)</title>
            <link>http://forums.innodb.com/read.php?7,333,333#msg-333</link>
            <description><![CDATA[ It would be great to have a 'sql' command like 'show tablespaces'  that  would provide detailed reports of space usage and free space available  of innodb tablespaces and files.]]></description>
            <dc:creator>cjgrillo</dc:creator>
            <category>Suggestion Box</category>
            <pubDate>Fri, 05 Dec 2008 00:06:37 +0200</pubDate>
        </item>
        <item>
            <guid>http://forums.innodb.com/read.php?7,154,154#msg-154</guid>
            <title>ibbackup needs better (PCRE) regex support, and --exclude flag (1 reply)</title>
            <link>http://forums.innodb.com/read.php?7,154,154#msg-154</link>
            <description><![CDATA[ I'm running a newly-purchased copy of innodb-hotbackup (ibbackup 3.0.0), and have found that the '--include' regex matching is insufficient to function for my needs.  <br />
<br />
I have a need when backing up databases/tables to be more specific than the POSIX EREG rules allow.  Particularly, I need the ability to exclude certain tables from backups; it is not possible to do so.  I need either a '--exclude' flag to the binary, or preferably an (optional) perl-conpatible regex (PREG) support so that lookahead/lookbehind assertions and other more complex matching is possible on the table/database names.<br />
<br />
One easy example is on a 24GB innodb database.  18GB of the data is stored in only 8 tables (of the 72 in this db), and I need to periodically not back up these large tables.  With a '--exclude' I could pass in '--exclude &quot;dbname\.history_*&quot;', and if PCRE regex support was added I could just do something like '--include &quot;(mysql|zabbix)/(?!history)&quot;'.<br />
<br />
This is a really necessary addition, I won't go into why you're using only POSIX EREG and have no equivalent --exclude, but with such expensive/niche commercial software, that's a big failing and an extremely easy addition.<br />
<br />
Thanks,<br />
<br />
/eli]]></description>
            <dc:creator>eli.stair</dc:creator>
            <category>Suggestion Box</category>
            <pubDate>Fri, 01 Aug 2008 02:51:18 +0300</pubDate>
        </item>
        <item>
            <guid>http://forums.innodb.com/read.php?7,117,117#msg-117</guid>
            <title>ibbackup wish list (1 reply)</title>
            <link>http://forums.innodb.com/read.php?7,117,117#msg-117</link>
            <description><![CDATA[ I want the following features:<br />
* backup writes to a socket rather than staging files locally<br />
* multiple threads used to read *.ibd files on backup<br />
* multiple threads used to copy in *.ibd files on restore<br />
<br />
Is there any chance we will get them? If you publish the source, I will add them.]]></description>
            <dc:creator>mdcallag</dc:creator>
            <category>Suggestion Box</category>
            <pubDate>Sat, 07 Jun 2008 03:16:01 +0300</pubDate>
        </item>
        <item>
            <guid>http://forums.innodb.com/read.php?7,94,94#msg-94</guid>
            <title>Using bloom filters to avoid disk hits in immutable tables? (2 replies)</title>
            <link>http://forums.innodb.com/read.php?7,94,94#msg-94</link>
            <description><![CDATA[ Let's say you have a table with 1M rows and let's say you have a 64bit int representing the primary key.<br />
<br />
This is roughly 8MB of data.<br />
<br />
If you have queries like:<br />
<br />
SELECT * FROM FOO WHERE ID = X;<br />
<br />
These can either use the index, or they could be ignored if you have a bloom filter.<br />
<br />
The filter would just include known records.<br />
<br />
This just used by MySQL to ignore the SQL and return an empty result set if there's no possibility that the record is on disk.<br />
<br />
With about an 800k bloom filter you can remove 97% of these checks.  <br />
<br />
The worse case scenario is that you end up falling through to disk for some of the false positives.<br />
<br />
Thoughts?]]></description>
            <dc:creator>burtonator</dc:creator>
            <category>Suggestion Box</category>
            <pubDate>Sat, 24 Jan 2009 03:07:18 +0200</pubDate>
        </item>
        <item>
            <guid>http://forums.innodb.com/read.php?7,64,64#msg-64</guid>
            <title>Defaults for innodb_row_format, key_block_size (1 reply)</title>
            <link>http://forums.innodb.com/read.php?7,64,64#msg-64</link>
            <description><![CDATA[ Suggestions from Venu Anuganti (http://venublog.com/).<br />
<br />
It will be nice if InnoDB team can extend the SHOW TABLE STATUS to show the compression details. Right now the Information schema.cmp will only lists general compression statistics but not on per table basis.<br />
<br />
Also, it will be added advantage if they can support the following two config variables:<br />
<br />
innodb_default_row_format (default compact or dynamic and making innodb_default_row_format=compact by default).<br />
innodb_default_key_block_size (optionally, and make sure it does not raise a warning on default row_format)]]></description>
            <dc:creator>KenJacobs</dc:creator>
            <category>Suggestion Box</category>
            <pubDate>Sat, 26 Apr 2008 22:24:34 +0300</pubDate>
        </item>
    </channel>
</rss>
