Hi Tim and all,<br><br><div class="gmail_quote">2013/4/8 Tim Landscheidt <span dir="ltr"><<a href="mailto:tim@tim-landscheidt.de" target="_blank">tim@tim-landscheidt.de</a>></span><br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
"Marc A. Pelletier" <<a href="mailto:marc@uberbox.org">marc@uberbox.org</a>> wrote:<br>
<br>
> [...]<br>
<br>
> The database replication is also well on its way; you can find the<br>
> current roadmap at:<br>
<br>
> <a href="https://wikitech.wikimedia.org/wiki/Tool_Labs/Database_plan" target="_blank">https://wikitech.wikimedia.org/wiki/Tool_Labs/Database_plan</a><br>
<br>
> [...]<br>
<br>
To quote from there:<br>
<br>
| Overview<br>
<br>
| * All public wikis will be replicated to the LabsDB servers,<br>
| with private user data redacted.<br>
<br>
| * First, data will be replicated to a special set of data-<br>
| base servers (PreLabsDBDBS) that use triggers to rewrite<br>
| or remove private data. They will write row based bin-<br>
| logs. Production shards will map 1:1 with mysql in-<br>
| stances, unlike on toolserver where some are combined via<br>
| a custom replication engine.<br>
<br>
| * Triggers will be created with the help of the redactatron<br>
| schema review tool.<br>
<br>
| * The actual labs databases will replicate from the above<br>
| mentioned databases. Users will access data via views<br>
| that only include reviewed tables and columns to ensure<br>
| that unreviewed tables (such as from a new extension)<br>
| aren't exposed without prior review.<br>
<br>
| * Replicated data will be stored on flash storage, while<br>
| each system will have a traditional disk array attached to<br>
| store labs project data. Users will be able to join<br>
| project tables against wiki tables, but only within the<br>
| current shard.<br>
<br>
| * The labs team will integrate these databases with labs,<br>
| automating database creation and access on a per-project<br>
| basis.<br>
<br>
This means that JOINs for example between wikis and Commons<br>
or Wikidata will not be possible. WTF? One of the stated<br>
goals of Tool Labs is "Provide a location for analytics<br>
work", so any changes here should /enhance/ the possibili-<br>
ties the Toolserver offers and not shrink them. This is BTW<br>
one of the top items on the "Needed Toolserver features"<br>
list.<br></blockquote></div><br>We will see into this and find out what is possible to do until when. You'll hear more about it soon.<br><br>Cheers, Silke<br clear="all"><br>-- <br>Silke Meyer<br>Internes IT-Management und Projektmanagement Toolserver<br>
<br>Wikimedia Deutschland e.V. | Obentrautstr. 72 | 10963 Berlin<br>Tel. (030) 219 158 260<br><br><a href="http://wikimedia.de" target="_blank">http://wikimedia.de</a><br><br>Wikimedia Deutschland - Gesellschaft zur Förderung Freien Wissens e.V. Eingetragen im Vereinsregister des Amtsgerichts Berlin-Charlottenburg unter der Nummer 23855 B. Als gemeinnützig anerkannt durch das Finanzamt für Körperschaften I Berlin, Steuernummer 27/681/51985.