BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Chicago
X-LIC-LOCATION:America/Chicago
BEGIN:DAYLIGHT
TZOFFSETFROM:-0600
TZOFFSETTO:-0500
TZNAME:CDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0500
TZOFFSETTO:-0600
TZNAME:CST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20181221T160904Z
LOCATION:C2/3/4 Ballroom
DTSTART;TZID=America/Chicago:20181115T083000
DTEND;TZID=America/Chicago:20181115T170000
UID:submissions.supercomputing.org_SC18_sess324_post234@linklings.com
SUMMARY:Distributed Adaptive Radix Tree for Efficient Metadata Search on H
 PC Systems
DESCRIPTION:Poster\nTech Program Reg Pass, Exhibits Reg Pass\n\nDistribute
 d Adaptive Radix Tree for Efficient Metadata Search on HPC Systems\n\nZhan
 g, Tang, Byna, Chen\n\nAffix-based search allows users to retrieve data wi
 thout the need to remember all relevant information precisely. While build
 ing an inverted index to facilitate efficient affix-based search is a comm
 on practice for standalone databases and desktop file systems, they are of
 ten insufficient for high-performance computing (HPC) systems due to the m
 assive amount of data and the distributed nature of the storage. In this p
 oster, we present Distributed Adaptive Radix Tree (DART) which enables sca
 lable and efficient affix-based search. DART maintains a balanced keyword 
 distribution and optimizes for excessive keyword requests dynamically at s
 cale. Our evaluation shows that compared with the “full string hashing” us
 ed by the commonly adopted DHT approach, DART achieves up to 55x throughpu
 t speedup for prefix and suffix search, and has a comparable throughput fo
 r exact and infix search. Also, DART maintains balanced keyword distributi
 on and alleviates excessive query workload on popular keywords.
URL:https://sc18.supercomputing.org/presentation/?id=post234&sess=sess324
END:VEVENT
END:VCALENDAR