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

»çÀÌÆ®¸Ê

Loading..

Please wait....

±¹³» ³í¹®Áö

Ȩ Ȩ > ¿¬±¸¹®Çå > ±¹³» ³í¹®Áö > Çѱ¹Á¤º¸°úÇÐȸ ³í¹®Áö > Á¤º¸°úÇÐȸ³í¹®Áö (Journal of KIISE)

Á¤º¸°úÇÐȸ³í¹®Áö (Journal of KIISE)

Current Result Document : 270 / 274

ÇѱÛÁ¦¸ñ(Korean Title) ½Ã°è¿­ access log data¸¦ ÀÌ¿ëÇÑ IT ÀÎÇÁ¶ó ÀÌ»ó¡ÈÄ °¨Áö ¾Ó»óºí ¸ðµ¨
¿µ¹®Á¦¸ñ(English Title) Ensemble Model for Detecting Abnormal Symptoms of IT Infrastructure using Time Series Access Log Data
ÀúÀÚ(Author) ±èÁ¤¿ø   ÃÖÈ£Áø   Jungwon Kim   Ho-Jin Choi  
¿ø¹®¼ö·Ïó(Citation) VOL 48 NO. 09 PP. 1035 ~ 1043 (2021. 09)
Çѱ۳»¿ë
(Korean Abstract)
´ë±Ô¸ð IT ¼­ºñ½º¸¦ ¿î¿µÇÏ´Â °÷¿¡¼­ ´ÜÁö ÇϳªÀÇ ½Ã½ºÅÛÀ» °ü¸®ÇÏ´Â °æ¿ì´Â ¸Å¿ì µå¹°´Ù. ¹°·Ð °üÁ¦¸¦ Àü´ãÇÏ´Â Á¶Á÷ÀÌ ÀÖ´Ù¸é ¼­ºñ½ºÀÇ ÀÌ»óÀ¯¹«¿¡ ´ëÇØ ¸ð´ÏÅ͸µÀÌ °¡´ÉÇÏ°ÚÁö¸¸, °üÁ¦ ´ã´çÀÚ´Â °¢ ¼­ºñ½ºÀÇ ¾÷¹« Áö½Ä°ú µµ¸ÞÀο¡ ´ëÇØ Àß ¾ËÁö ¸øÇϱ⠶§¹®¿¡, ƯÁ¤ ¼­ºñ½ºÀÇ ºñÁ¤»ó ¿©ºÎ¸¦ ÆÇ´ÜÇϱ⠾î·Á¿î °ÍÀÌ »ç½ÇÀÌ´Ù. µû¶ó¼­ °¢ ¼­ºñ½º¸¶´ÙÀÇ Æ¯¼ºÀ» ºÐ¼®ÇÏ°í ÆÐÅÏÀ» ÇнÀÇÏ¿© À̻󿩺θ¦ ÆÇ´ÜÇϴ ŽÁö ¸ðµ¨ÀÇ needs°¡ ³ª³¯ÀÌ Áõ°¡ÇÏ°í ÀÖ´Ù. º» ¿¬±¸¿¡¼­´Â À¥¼­¹öÀÇ access log¿¡ ±â·ÏµÇ¾î ÀÖ´Â ½Ã°è¿­ µ¥ÀÌÅ͸¦ ÀÌ¿ëÇÏ¿©, ±âÁ¸ ½ºÆåÆ®·³ ÀÜÂ÷ ¹æ½ÄÀÇ ¸ðµ¨ÀÌ ½Ç½Ã°£À¸·Î ÀÌ»ó¡Èĸ¦ ŽÁöÇÒ ¼ö ÀÖÀ»Áö¿¡ ´ëÇØ »ìÆ캸°í, ½Ç½Ã°£ ŽÁö°¡ ¾î·Á¿î ¹®Á¦¸¦ ÇØ°áÇϱâ À§ÇØ ´ÙÇ×ȸ±Í¸ðµ¨°ú ¾Ó»óºíÇÑ ¸ðµ¨À» Á¦½ÃÇÔÀ¸·Î½á, Àå¾Ö»óȲÀÌ ¹ß»ýÇϱâ Àü¿¡ ºü¸¥ ´ëó¸¦ ÇÒ ¼ö ÀÖµµ·Ï ¸ðµ¨À» ±¸ÇöÇÏ¿´´Ù. ±× °á°ú ½Ã½ºÅÛ Àå¾Ö°¡ ¹ß»ýÇϱâ Àü¿¡ ÀÌ»ó¡Èĸ¦ °¨ÁöÇÏ¿© ¼±Á¦´ëÀÀÀ» ÇÒ ¼ö ÀÖÀ½À» È®ÀÎÇÒ ¼ö ÀÖ¾ú´Ù.
¿µ¹®³»¿ë
(English Abstract)
When operating large-scale IT services, multiple systems need to be managed for the detection of abnormalities. Since the monitoring and control personnel can hardly have all the domain knowledge required for these systems and services, there have been increasing needs for automated models that detect abnormalities by analyzing the characteristics of each service and learning patterns. In this experiment, we use the time-series data in the access log of the web server to examine the capability of the existing spectrum residual method model in detecting the anomalies in real-time, and propose an improved detection model which can respond more quickly to an abnormal situation. Our experiment showed that the proposed model was able to predict abnormal symptoms before actual failure occurs, and to respond in advance.
Å°¿öµå(Keyword) ÀÌ»ó¡ÈÄ°¨Áö   ½Ã°è¿­µ¥ÀÌÅÍ   ½ºÆåÆ®·³ÀÜÂ÷   ´ÙÇ×ȸ±Í   ±â°èÇнÀ   ¾Ó»óºí¸ðµ¨   abnormality detection   time series data   spectrum residual   polynomial regression   machine learning   ensemble model  
ÆÄÀÏ÷ºÎ PDF ´Ù¿î·Îµå