Not exactly, transaction information is stored in disk to restore it in case you stop a data node and restart it again.<br><br>The idea is that there are several data nodes storing replicated data and some&nbsp;of&nbsp;them can fail as long as there are nodes available (redundancy can be specified in the cluster parameters (noOfReplicas,...)). When the&nbsp;failed node come online again it retrieves the data from the surviving nodes so the data survives to node failures.
<br><br>You can start a cluster backup from the management node to effectively write the cluster information on disk so you can afterwards recover the cluster.<br><br>From mysql 5.1 you can use ndb engine and configure it to store the data on the disk instead of storing it on memory. However keys are still required to be stored on memory.
<br><br>Summarising,<br>Samuel.<br><br><div><span class="gmail_quote">2007/11/15, Iñaki Baz Castillo &lt;<a href="mailto:ibc@in.ilimit.es">ibc@in.ilimit.es</a>&gt;:</span><blockquote class="gmail_quote" style="margin-top: 0; margin-right: 0; margin-bottom: 0; margin-left: 0; margin-left: 0.80ex; border-left-color: #cccccc; border-left-width: 1px; border-left-style: solid; padding-left: 1ex">
El Thursday 15 November 2007 08:13:53 Klaus Darilion escribió:<br><br>&gt; &gt; The basic conclusion: NDB (mysql cluster) is the fastest engine because<br>&gt; &gt; it is a main-memory engine and has advanced locking like InnoDB.
<br>&gt;<br>&gt; So if the cluster fails, all the data is gone?<br><br>No experience with this, but I assume data is saved to disk periodically,<br>isn&#39;t?<br><br>--<br>Iñaki Baz Castillo<br><a href="mailto:ibc@in.ilimit.es">
ibc@in.ilimit.es</a><br><br>_______________________________________________<br>Users mailing list<br><a href="mailto:Users@lists.openser.org">Users@lists.openser.org</a><br><a href="http://lists.openser.org/cgi-bin/mailman/listinfo/users">
http://lists.openser.org/cgi-bin/mailman/listinfo/users</a><br></blockquote></div><br>