• Àüü
  • ÀüÀÚ/Àü±â
  • Åë½Å
  • ÄÄÇ»ÅÍ
´Ý±â

»çÀÌÆ®¸Ê

Loading..

Please wait....

±¹³» ÇÐȸÁö

Ȩ Ȩ > ¿¬±¸¹®Çå > ±¹³» ÇÐȸÁö > µ¥ÀÌÅͺ£À̽º ¿¬±¸È¸Áö(SIGDB)

µ¥ÀÌÅͺ£À̽º ¿¬±¸È¸Áö(SIGDB)

Current Result Document :

ÇѱÛÁ¦¸ñ(Korean Title) µ¥ÀÌÅÍ º´·Ä¼ºÀ» È°¿ëÇÑ ·Î±× ÆÄÀÏ À籸Ãà ºÎÇÏ °³¼± ¿¬±¸
¿µ¹®Á¦¸ñ(English Title) A Study on Improvement of Log File Reconstruction Overhead Utilizing Data Parallelism
ÀúÀÚ(Author) ¼­ÁÖ¿¬   ¼ºÇѽ   ÃÖ¿ø±â   ¹Ú»óÇö   Juyeon Seo   Hanseung Sung   Won Gi Choi   Sanghyun Park  
¿ø¹®¼ö·Ïó(Citation) VOL 36 NO. 03 PP. 0056 ~ 0075 (2020. 12)
Çѱ۳»¿ë
(Korean Abstract)
Èֹ߼º ÀåÄ¡ÀÎ DRAMÀ» ÁÖ ÀúÀå¼Ò·Î »ç¿ëÇÏ´Â Àθ޸𸮠µ¥ÀÌÅͺ£À̽ºÀÎ Redis´Â ½Å¼ÓÇÑ µ¥ÀÌÅÍ Ã³¸®ÀÇ ÀåÁ¡À» Áö³æÁö¸¸, µ¥ÀÌÅÍ À¯½Ç À§ÇèÀÌ Å©´Ù. Redis´Â µ¥ÀÌÅÍ À¯½Ç ¹æÁö¸¦ À§ÇØ, ¼öÇàÇÑ ¸í·É¾î¸¦ ·Î±× ÆÄÀÏ¿¡ µ¡ºÙ¿© ±â·ÏÇÏ´Â AOF ±â¹ýÀ» Á¦°øÇÑ´Ù. ÇÏÁö¸¸ °è¼ÓµÇ´Â ±â·ÏÀ¸·Î ÀÎÇÏ¿© ·Î±× ÆÄÀÏÀÌ ±â¾à ¾øÀÌ Áõ°¡ÇÏ´Â Çö»óÀÌ ¹ß»ýÇÑ´Ù. Redis´Â ·Î±× ÆÄÀÏ À籸Ãà ±â¹ýÀ» Á¦°øÇÏ¿© ·Î±× ÆÄÀÏ ±ÞÁõÀ» ¹æÁöÇÏÁö¸¸, ºÎ°¡ÀûÀÎ ¸Þ¸ð¸® »ç¿ë°ú µ¥ÀÌÅÍ Ã³¸® Áö¿¬ÀÌ ¿ä±¸µÇ¾î ¼º´É ¹× ÀÚ¿ø Ãø¸é¿¡¼­ ºÎ´ãÀÌ °¡ÁߵȴÙ. º» ¿¬±¸¿¡¼­´Â ·Î±× ÆÄÀÏ À籸 Ãà ½Ã ¹ß»ýÇÏ´Â ºÎÇϸ¦ °³¼±Çϱâ À§ÇØ µ¥ÀÌÅÍ º´·Ä¼ºÀ» È°¿ëÇÑ ·Î±× ÆÄÀÏ À籸Ãà ±â¹ýÀ» Á¦¾ÈÇÑ´Ù. ¶ÇÇÑ, Àç ±¸Ãà ºÎÇÏÀÇ ÁÖ¿ä ¿øÀÎÀÌ µÇ´Â AOF Rewrite ¹öÆÛ ¾øÀÌ À籸Ãà ÀÛ¾÷À» ¼öÇàÇÏ¿©, ¸Þ¸ð¸® »ç¿ë·®°ú µ¥ÀÌÅÍ Ã³¸® ¼º´ÉÀ» Çâ»ó½ÃŲ´Ù. Á¦¾ÈÇÏ´Â ¸ðµ¨Àº memtier-benchmark¸¦ ÅëÇØ Æò°¡µÇ¾úÀ¸¸ç, À籸Ãà ºÎÇÏ°¡ Å©°Ô °³¼±µÇ¾úÀ½À» È®ÀÎÇÏ¿´´Ù
¿µ¹®³»¿ë
(English Abstract)
Redis, an in-memory database that uses DRAM which is a volatile device as its main repository, has the advantage of rapid data processing but the risk of data loss is high. To prevent the data loss, Redis provides AOF method that appends log records for performed commands to the log file. However, there is a constant increase in the log file due to continuous records. Although Redis provides log file reconstruction methods to prevent sudden increase in the log file, additional memory usage and latency in data processing are required, adding to the burden in terms of performance and resources. In this research, we propose the log file reconstruction method using data parallelism to improve the overhead occurred in the log file reconstruction. Also, we perform reconstruction without AOF Rewrite buffer that is the major cause of reconstruction overhead through our method, so the memory usage and data processing performance are improved. The proposed model is evaluated via memtier-benchmark and we confirmed the reconstruction overhead has been considerably improved.
Å°¿öµå(Keyword) Log File Reconstruction   Data Parallelism   In-memory Database   Redis   ·Î±× ÆÄÀÏ À籸Ãà   µ¥ÀÌÅÍ º´·Ä¼º   Àθ޸𸮠µ¥ÀÌÅͺ£À̽º   ·¹µð½º  
ÆÄÀÏ÷ºÎ PDF ´Ù¿î·Îµå