• 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 : 9 / 14 ÀÌÀü°Ç ÀÌÀü°Ç   ´ÙÀ½°Ç ´ÙÀ½°Ç

ÇѱÛÁ¦¸ñ(Korean Title) °Ë»ö¿£ÁøÀ» À§ÇÑ ¸ÖƼÇÁ·Î¼¼½º/¸ÖƼ¾²·¹µå ¸ðµ¨ÀÇ ¼³°è ¹× ±¸Çö
¿µ¹®Á¦¸ñ(English Title) Design and Implementation of a Multi-Process/Multi-Thread Model for the ODYSSEUS/Parallel-OOSQL Parallel DBMS/Search Engine
ÀúÀÚ(Author) À±Å¼·   ¿©¿¬¹Ì   ÀÌÀÏ¿±   Ȳ±Ô¿µ   Tae-Seob Yun   Yeon-Mi Yeo   Ilyeop Yi   Kyu-Young Whang  
¿ø¹®¼ö·Ïó(Citation) VOL 33 NO. 02 PP. 0089 ~ 0112 (2017. 08)
Çѱ۳»¿ë
(Korean Abstract)
Google, Yahoo!¿Í °°Àº °Ë»ö ¿£ÁøÀº ¼ö¹é¾ï °ÇÀÇ À¥ ¹®¼­¸¦ »öÀÎÇÏ°í, ´Ù¼öÀÇ ¸Ó½Åµé¿¡¼­ ÁúÀǸ¦ º´·Ä·Î ó¸®ÇÑ´Ù. º´·Ä °Ë»ö ¿£ÁøÀº ´ë¿ë·® µ¥ÀÌÅ͸¦ ÀúÀåÇÏ°í ¼ö¸¹Àº »ç¿ëÀڵ鿡°Ô ¼­ºñ½º¸¦ Á¦°øÇØ¾ß ÇϹǷΠȮÀ强ÀÌ ÁÁ¾Æ¾ß ÇÑ´Ù. ¸ÖƼÇÁ·Î¼¼½º/½Ì±Û¾²·¹µå ¸ðµ¨À» »ç¿ëÇÏ´Â °Ë»ö ¿£ÁøÀº ´Ù¼öÀÇ ÇÁ·Î¼¼½º¿¡¼­ ÁúÀǸ¦ º´·Ä·Î ó¸®Çϱâ À§ÇØ ¸¹Àº ¾çÀÇ ¸Þ¸ð¸®¸¦ ÇÊ¿ä·Î Çϱ⠶§¹®¿¡ È®À强ÀÌ ÁÁÁö ¾Ê´Ù. ¹Ý¸é, ¸ÖƼÇÁ·Î¼¼½º/¸ÖƼ¾²·¹µå ¸ðµ¨À» »ç¿ëÇÒ °æ¿ì, ÇϳªÀÇ ÇÁ·Î¼¼½º°¡ ´Ù¼öÀÇ ¾²·¹µå¸¦ °¡Áü¿¡ µû¶ó ½Ã½ºÅÛ¿¡¼­ »ý¼ºµÇ´Â ÇÁ·Î¼¼½º °³¼ö°¡ °¨¼ÒÇÏ¿© ¸ÖƼ¾²·¹µå/½Ì±Û¾²·¹µå ¸ðµ¨¿¡ ºñÇØ È®À强ÀÌ ÁÁ´Ù. º» ³í¹®¿¡¼­´Â º´·Ä DBMS/°Ë»ö¿£ÁøÀÎ ¿Àµð¼¼¿ì½º/Parallel-OOSQLÀ» À§ÇÑ ¸ÖƼÇÁ·Î¼¼½º/¸ÖƼ¾²·¹µå ¸ðµ¨À» ¼³°è ¹× ±¸ÇöÇÑ´Ù. Á¦¾ÈµÈ ¸ðµ¨Àº ½Ã½ºÅÛ¿¡ »ý¼ºµÇ´Â ÇÁ·Î¼¼½º °³¼ö, ¸Þ¸ð¸® »ç¿ë·®, ±×¸®°í ÄÁÅؽºÆ® ½ºÀ§Äª ¿À¹öÇìµå¸¦ Å©°Ô °¨¼Ò½ÃŲ´Ù. ¶ÇÇÑ, ÇÁ·Î¼¼½º/¾²·¹µåµéÀ» È¿°úÀûÀ¸·Î °ü¸®ÇÏ°í, ÁúÀÇ Ã³¸® ¼Óµµ¸¦ Çâ»ó½ÃÅ°±â À§ÇÑ ¹æ¹ýÀ» Á¦¾ÈÇÑ´Ù. ½ÇÇèÀ» ÅëÇؼ­´Â Á¦¾ÈµÈ ¸ðµ¨ÀÌ ¸ÖƼ¾²·¹µå/½Ì±Û¾²·¹µå ¸ðµ¨º¸´Ù È®À强ÀÌ ÁÁÀ½À» º¸ÀδÙ. ½ÇÇè °á°ú, Á¦¾ÈµÈ ¸ðµ¨Àº ¸Þ¸ð¸® »ç¿ë·®ÀÌ 69%, ÁúÀÇ Ã³¸® ¼Óµµ°¡ 19%ÀÇ Çâ»óµÇ¾ú´Ù.
¿µ¹®³»¿ë
(English Abstract)
Search engines such as Google and Yahoo! index tens of billions of web pages and process queries in massive-scale parallelism using multiple machines. Scalability of parallel search engines is very important because their engines need to support a huge volume of data and a large number of users. Search engines using multi-process/single-thread model are not scalable because they consume large amount of memory resources to process queries in parallel in multiple processes. On the other hand, multi-process/multi-thread models are more scalable than multi-process/single-thread models because their processes consists of multiple threads, which reduces the number of processes. In this paper, we design and implement a multi-process/multi-thread model for ODYSSEUS/Parallel-OOSQL parallel DBMS/search engine. The proposed model significantly decreases the number of processes of ODYSSEUS/Parallel-OOSQL and reduces the amount of memory usage and context switching overhead. Besides, we propose an effective method for maintaining processes/threads and increasing the query processing speed. We also conduct experiments to show that the proposed multi-process/multi-thread model is more scalable than the multi-process/single-thread model. The experimental results show that the proposed model outperforms up to 69% in memory usage and 19% in query processing speed.
Å°¿öµå(Keyword) º´·Ä DBMS   º´·Ä °Ë»ö ¿£Áø   ¸ÖƼÇÁ·Î¼¼½º   ¸ÖƼ¾²·¹µå   parallel DBMS   parallel search engine   multi-process   multi-thread  
ÆÄÀÏ÷ºÎ PDF ´Ù¿î·Îµå