Á¤º¸°úÇÐȸ³í¹®Áö (Journal of KIISE)
Current Result Document :
ÇѱÛÁ¦¸ñ(Korean Title) |
´ë±âÇà·Ä ¸ðµ¨À» È°¿ëÇÑ °ÈÇнÀ ±â¹Ý ÆÄµå ¿ÀÅ佺ÄÉÀϸµ ±â¹ý |
¿µ¹®Á¦¸ñ(English Title) |
Reinforcement Learning-based Pod Autoscaling Technique Using the Queueing Model |
ÀúÀÚ(Author) |
Àå¿ëÇö
À¯Çåâ
ÀÌÀº¿µ
Yonghyeon Jang
Heonchang Yu
Eunyoung Lee
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 49 NO. 02 PP. 0106 ~ 0119 (2022. 02) |
Çѱ۳»¿ë (Korean Abstract) |
Ãֱ٠ȯ°æ º¯È¿¡ ÀûÀÀÀûÀÌ°í ¸ñÀû¿¡ ¸Â´Â ÃÖÀûÀÇ ¿ÀÅ佺ÄÉÀϸµ Á¤Ã¥À» »ç¿ëÇϱâ À§ÇØ °ÈÇнÀ ±â¹Ý ¿ÀÅ佺ÄÉÀϸµ Á¤Ã¥¿¡ ´ëÇÑ ¿¬±¸°¡ ÀÌ·ç¾îÁö°í ÀÖ´Ù. ±×·¯³ª °ÈÇнÀ ±â¹Ý ¿ÀÅ佺ÄÉÀϸµ Á¤Ã¥À» ÇнÀÇÏ°í, °¢°¢ÀÇ °ÈÇнÀ ±â¹Ý ¿ÀÅ佺ÄÉÀϸµ Á¤Ã¥ °£ÀÇ ¼º´É ºñ±³¸¦ ¼öÇàÇÏ´Â °úÁ¤¿¡¼ ¸¹Àº ½Ã°£°ú ÀÚ¿øÀÌ ¿ä±¸µÈ´Ù´Â ¹®Á¦°¡ ¹ß»ýÇÑ´Ù. º» ³í¹®¿¡¼´Â ´ë±âÇà·Ä ¸ðµ¨ ±â¹Ý ½Ã¹Ä·¹ÀÌ¼Ç ±â¹ýÀ» Á¦¾ÈÇÏ¿© ¿ÀÅ佺ÄÉÀϸµ Á¤Ã¥ °£ÀÇ ¼º´É ºñ±³¸¦ ½Ã¹Ä·¹À̼ÇÀ» ÅëÇØ ¼öÇàÇÒ ¼ö ÀÖ°Ô ÇÏ°í, ½Ã¹Ä·¹ÀÌ¼Ç ½ÇÇèÀ» ÅëÇØ ¿©·¯ °ÈÇнÀ ±â¹Ý ÆÄµå ¿ÀÅ佺ÄÉÀϸµ ±â¹ýÀ» ºñ±³ÇÑ´Ù. |
¿µ¹®³»¿ë (English Abstract) |
Recently, studies on reinforcement learning-based autoscaling policies have been conducted in order to use optimal autoscaling policies that are adaptive to environmental changes and fit the purpose. However, there is a problem that much time and many resources are required in the process of training the reinforcement learning-based autoscaling policy and comparing the performance between each reinforcement learning-based autoscaling policy. In this study, we proposed a queueing model-based simulation technique, which enables performance comparison between autoscaling policies to be performed through simulation, and compared several reinforcement learning-based pod autoscaling techniques through simulation experiments. |
Å°¿öµå(Keyword) |
Äí¹ö³×Ƽ½º
°ÈÇнÀ
¿ÀÅ佺ÄÉÀϸµ
´ë±âÇà·Ä ¸ðµ¨
kubernetes
reinforcement learning
autoscaling
queueing model
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|