http://wiki.define-technology.com/mediawiki-1.35.0/index.php?action=history&feed=atom&title=Luster_Intel%3A_Notes_from_training 2026-05-04T22:51:28Z Revision history for this page on the wiki MediaWiki 1.35.0 http://wiki.define-technology.com/mediawiki-1.35.0/index.php?title=Luster_Intel:_Notes_from_training&diff=3907&oldid=prev 2014-01-29T11:15:02Z

<p>Created page with &quot;== Luster Training, Swindon, January 2014 == === What is lustre === * multi vendor global parallel file system * Parallel, Shared, Multi-Vendor, Open source, Scalable, Provi...&quot;</p> <p><b>New page</b></p><div>== Luster Training, Swindon, January 2014 ==<br /> <br /> === What is lustre ===<br /> <br /> * multi vendor global parallel file system<br /> * Parallel, Shared, Multi-Vendor, Open source, Scalable, Provided by intel<br /> * Not currently a windows client - working on a samba client that will produce good, but not great results<br /> <br /> * scale out = more servers = more bandwidth<br /> * scale up = more targets = more storage<br /> <br /> === Meta Data Servers ===<br /> <br /> * turn on HT to make use of as many threads as possible<br /> * 2.5 is going to focus on the meta date server - looking to turn it into a cluster<br /> * Can be set up with HA failover<br /> <br /> == Object servers ==<br /> <br /> * Suggest setting up with pairs to allow for failover<br /> * <br /> <br /> == Main Componants==<br /> <br /> * Management Target - Luster mount service, intal point on contact<br /> * Metadata Target<br /> * Object Storage Targets<br /> <br /> == General Recommendations ==<br /> <br /> Managment - Barebones install on its on system (avoid sharing databases etc)</div>

Michael