• 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 ½ºÆ®¸²ÀÇ »ùÇøµ ¹× ÇÊÅ͸µÀ» À§ÇÑ °í¼Ó ºÐ»ê ó¸® Ç÷§Æû
¿µ¹®Á¦¸ñ(English Title) High-speed Distributed Processing Platform for Sampling and Filtering Data Streams
ÀúÀÚ(Author) ±æ¸í¼±   ¹®¾ç¼¼   ÃÖÇüÁø   Myeong-Seon Gil   Yang-Sae Moon   Hyung-Jin Choi  
¿ø¹®¼ö·Ïó(Citation) VOL 37 NO. 03 PP. 0016 ~ 0028 (2021. 12)
Çѱ۳»¿ë
(Korean Abstract)
º» ³í¹®¿¡¼­´Â µ¥ÀÌÅÍ Ç°Áú °ü¸®ÀÇ ´ëÇ¥ ±â¼úÀÎ µ¥ÀÌÅÍ Á¤Á¦(purification)¸¦ ½ºÆ®¸² ȯ°æ¿¡ Àû¿ëÇÒ ¼ö ÀÖ´Â »õ·Î¿î ºÐ»ê ó¸® Ç÷§ÆûÀ» Á¦¾ÈÇÑ´Ù. À̸¦ À§ÇØ, ¸ÕÀú ±âÁ¸ Á¤Á¦ ±â¼úÀÇ ¼¼ °¡Áö ¹®Á¦Á¡À» ºÐ¼®ÇÑ´Ù. ±×¸®°í, °¢ ¹®Á¦Á¡ÀÇ ÇØ°á ¹æ¾ÈÀ¸·Î ¿ÀǼҽº ÇÁ·ÎÁ§Æ® ±â¹ÝÀÇ »õ·Î¿î °í¼Ó ½ºÆ®¸² ó¸® Ç÷§ÆûÀ» Á¦½ÃÇÑ´Ù. Á¦¾È Ç÷§ÆûÀº µ¥ÀÌÅÍ ½ºÆ®¸² ó¸® ¿£Áø, Á¤Á¦ ¶óÀ̺귯¸®, Ç÷£ ¸Å´ÏÀú, °øÀ¯ ½ºÅ丮Áö·Î ±¸¼ºµÇ¸ç, Apache Storm ¹× Apache Kafka¸¦ ±â¹ÝÀ¸·Î ¼³°èÇÑ´Ù. ÇØ´ç ¿ä¼ÒµéÀº ½ºÆ®¸² ó¸® ¼Óµµ ¹× 󸮷®ÀÌ ¸Å¿ì Áß¿äÇÑ ¼º´É ÁöÇ¥·Î, º» ³í¹®¿¡¼­´Â RDMA(Remote Direct Memory Access)¸¦ È°¿ëÇÏ¿© ¼º´ÉÀ» Çâ»ó½ÃŲ´Ù. Á¦¾È Ç÷§ÆûÀÇ ¼º´É Æò°¡¿¡´Â ÃÑ ¾ÆÈ© ´ë ³ëµå·Î ±¸¼ºµÈ ºÐ»ê Ŭ·¯½ºÅÍ È¯°æÀ» »ç¿ëÇϸç, À̸¦ ±â¹ÝÀ¸·Î °¢ ÄÄÆ÷³ÍÆ®¸¦ ±¸Çö ¹× Æò°¡ÇÏ¿´´Ù. ½ÇÇèÀ» ÅëÇØ Á¦¾È Ç÷§ÆûÀÇ ½ºÆ®¸² 󸮷®ÀÌ ±âÁ¸ ÀÌ´õ³Ý ȯ°æ ´ëºñ Æò±Õ 28¹è ÀÌ»ó, 󸮽ð£Àº Æò±Õ 2,473¹è ÀÌ»ó Çâ»óµÇ¾úÀ½À» È®ÀÎÇÏ¿´´Ù. °á°úÀûÀ¸·Î, º» ³í¹®¿¡¼­ Á¦¾ÈÇÑ ½ºÆ®¸² Á¤Á¦ Ç÷§ÆûÀº ÃÊ°í¼Ó, ´ë¿ë·®À¸·Î ¹ß»ýÇÏ´Â ½ºÆ®¸²ÀÇ È¿À²ÀûÀÎ Á¤Á¦¸¦ Áö¿øÇÏ´Â ÃÖÃÊÀÇ ÅëÇÕ Ç÷§ÆûÀ̶ó ÇÒ ¼ö ÀÖ´Ù.
¿µ¹®³»¿ë
(English Abstract)
In this paper, we propose a novel intelligent platform that applies data purification, a representative technology of data quality management, to the data stream environment. First, we analyze three problems of the existing purification technology: Unsuitability on high-speed stream environment, Lack of stream-based purification methods, and High-difficulty in utilizing purification technologies. Next, to solve those problems, we derive a new high-speed stream processing platform based on open-source projects. The proposed platform consists of Data Stream Processing Engine, Purification Library, Plan Manager, and Shared Storage, and we implement our platform based on Apache Storm and Apache Kafka. Stream processing speed and throughput of these elements are very important performance measures. Thus, we address a performance improvement method using RDMA-Storm (Remote Direct Memory Access based Storm) to increase these performance measures. Through extensive experiments, we showed that the throughput of the proposed platform improved by over 28 times and 2,473 times of processing time compared to the existing Ethernet environment. To our best knowledge, the proposed stream purification platform is the first integrated platform that supports stable purification on an ultra-high speed and big data stream environment.
Å°¿öµå(Keyword) µ¥ÀÌÅÍ ½ºÆ®¸²   µ¥ÀÌÅÍ Á¤Á¦   °í¼Ó ½ºÆ®¸² 󸮠  ºÐ»ê ½ºÆ®¸² 󸮠  ¾ÆÆÄÄ¡ ½ºÅè   Data stream   Data purification   High-speed stream processing   Distributed stream processing   Apache Storm  
ÆÄÀÏ÷ºÎ PDF ´Ù¿î·Îµå