BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20191206T142452Z
LOCATION:Analog 1\, 2
DTSTART;TZID=Europe/Stockholm:20180625T135700
DTEND;TZID=Europe/Stockholm:20180625T140100
UID:isc_hpc_ISC High Performance 2018_sess103_phd104@linklings.com
SUMMARY:(PhD15) HasFS: A File System for NVM-based Hybrid Storage Architec
 ture
DESCRIPTION:PhD Forum\n\n(PhD15) HasFS: A File System for NVM-based Hybrid
  Storage Architecture\n\nLiu, Lu, Chen, Du, Xiao\n\nIn traditional DRAM-ba
 sed file systems (such as EXT4), file systems need to periodically synchro
 nize the dirty data on the DRAM to the disk to protect against data crash 
 because DRAM is a volatile media. Furthermore, some unpredictable system f
 ailures may break file system consistency. Therefore, many file systems us
 e journaling to guarantee the file system consistency. In our observations
 , periodic disk synchronization and after-crash consistency guarantee (jou
 rnaling) will significantly degrade file system performance at many worklo
 ads. So, we leverage NVM to tackle these two performance bottlenecks. \n
  Currently, non-volatile main memory (NVM) brings some new motivations to 
 file system optimization.  We leverage NVM to tackle these two performance
  bottlenecks. We present HasFS, a file system designed for the DRAM-NVM-DI
 SK architecture (Hybrid Architecture for Storage). HasFS extends the main 
 memory with NVM and treats NVM as a persistent page cache to eliminate the
  periodic disk synchronization overhead of dirty data. In addition, tradit
 ional journaling is inefficient on HasFS architecture because it depends o
 n the write ordering. Therefore, we designed an NVM-friendly, non-ordering
  consistency mechanism based on the hybrid storage architecture. This mech
 anism can provide strong consistency guarantee with low overhead. Our empi
 rical evaluation demonstrates that HasFS outperforms DRAM-based file syste
 ms at many workloads, even through NVM is slower than DRAM.\n\nPasses: Con
 ference Pass, Exhibition Pass, File Systems, Heterogeneous Systems, Storag
 e Technologies\n\nTag: Conference Pass, Exhibition Pass, File Systems, Het
 erogeneous Systems, Storage Technologies
URL:https://2018.isc-program.com/presentation/?id=phd104&sess=sess103
END:VEVENT
END:VCALENDAR