• 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¸°úÇÐȸ ÄÄÇ»ÆÃÀÇ ½ÇÁ¦ ³í¹®Áö (KIISE Transactions on Computing Practices)

Á¤º¸°úÇÐȸ ÄÄÇ»ÆÃÀÇ ½ÇÁ¦ ³í¹®Áö (KIISE Transactions on Computing Practices)

Current Result Document :

ÇѱÛÁ¦¸ñ(Korean Title) ICT ÀÀ¿ë ¼­ºñ½º Áö´ÉÈ­¸¦ Áö¿øÇϴ Ŭ¶ó¿ìµå-³×ÀÌƼºê ±â¹Ý SmartX AI ÄÄÇ»Æà Ŭ·¯½ºÅÍ ¼³°è ¹× °ËÁõ
¿µ¹®Á¦¸ñ(English Title) Design and Evaluation of Cloud-Native-based SmartX AI Computing Cluster Supporting AI-enabled Services
ÀúÀÚ(Author) ±ÇÁøö   ±è³²°ï   ±èÁ¾¿ø   JinCheol Kwon   Namgon Lucas Kim   JongWon Kim  
¿ø¹®¼ö·Ïó(Citation) VOL 25 NO. 12 PP. 0571 ~ 0584 (2019. 12)
Çѱ۳»¿ë
(Korean Abstract)
ÃÖ±Ù ¸¶ÀÌÅ©·Î ¼­ºñ½º ±¸Á¶(Micro Service Architecture: MSA)¸¦ Àû¿ëÇÑ Áö´ÉÇü ICT ÀÀ¿ë ¼­ºñ½º°¡ Å©°Ô Áõ°¡Çϸ鼭, À̵é°ú È¿°úÀû(Effectively)ÀÌ°í À¯¿¬ÇÏ°Ô(Flexibly) ¿¬°èµÉ ¼ö ÀÖ´Â AI ÄÄÇ»Æà ÀÚ¿øµé¿¡ ´ëÇÑ ¿ä±¸ ¶ÇÇÑ Å©°Ô Áõ°¡ÇÏ°í ÀÖ´Ù. º» ³í¹®¿¡¼­´Â »ó±âÇÑ AI ¼ö¿ä¿¡ È¿°úÀûÀ¸·Î ´ëÀÀÇÒ ¼ö Àִ Ŭ¶ó¿ìµå-³×ÀÌƼºê(Cloud-native) ±â¹ÝÀÇ SmartX AI ÄÄÇ»Æà Ŭ·¯½ºÅÍ(SmartX AI Computing Cluster)ÀÇ °³³äÀ» Á¦¾ÈÇÑ´Ù. ¿©±â¼­ Ŭ¶ó¿ìµå-³×ÀÌƼºê ÄÄÇ»ÆÃÀÇ ÇÙ½É ±â¼úÀÎ ÄÁÅ×ÀÌ³Ê °¡»óÈ­¿Í ÄÁÅ×ÀÌ³Ê ¿ÀÄɽºÆ®·¹ÀÌ¼Ç ±â¼úÀº ºÐ»ê ºòµ¥ÀÌÅÍ/¸Ó½Å·¯´× ¿öÅ©·Îµå¸¦ ´õ¿í À¯¿¬ÇÏ°í µ¿ÀûÀ¸·Î ¹èÆ÷/¿î¿ëÇÒ ¼ö ÀÖµµ·Ï µµ¿òÀ» ÁØ´Ù. Á¦¾ÈÇϴ Ŭ·¯½ºÅÍÀÇ ±¸¼ºÀº AI ÄÄÇ»Æÿ¡ ƯȭµÈ °í¼º´É Çϵå¿þ¾î¿Í ¿ÀǼҽº ¼ÒÇÁÆ®¿þ¾î¸¦ Áß½ÉÀ¸·Î ÇÑ´Ù. º» ³í¹®¿¡¼­´Â Á¦½ÃÇÑ Å¬·¯½ºÅÍÀÇ °³³ä¿¡ µû¶ó ½ÇÁ¦ Ŭ·¯½ºÅ͸¦ ±¸ÃàÇϱâ À§ÇØ ÇÊ¿äÇÑ ±¸Ã¼ÀûÀÎ Çϵå¿þ¾îÀÇ ±¸¼º°ú ¼ÒÇÁÆ®¿þ¾îÀÇ ½ºÅÃÀ» ¼³°èÇÏ°í Á¦½ÃÇÑ´Ù. ±×¸®°í Á¦½ÃÇÑ ¼³°è¿¡ µû¶ó ¼Ò±Ô¸ð Ŭ·¯½ºÅ͸¦ ½ÃÇèÀûÀ¸·Î ±¸ÃàÇÏ¿© º¸°í, ±¸ÃàµÈ Ŭ·¯½ºÅÍÀÇ ±â´Éµé°ú È°¿ë ¹æ¾È¿¡ ´ëÇØ »ìÆ캻´Ù. ¶ÇÇÑ ½ÇÁ¦ ºòµ¥ÀÌÅÍ ¹× ºÐ»ê ¸Ó½Å·¯´× Æ®·¹ÀÌ´× ¿öÅ©·Îµå ¿î¿ë ½ÇÇèÀ» ÅëÇØ ½ºÅ丮Áö¿Í ³×Æ®¿öÅ· Ãø¸é¿¡¼­ ¹ß»ýÇÏ´Â º´¸ñ(Bottleneck Points)À» µµÃâÇÏ¿© º¸°í, À̸¦ °³¼±Çϱâ À§ÇÑ ¼ÒÇÁÆ®¿þ¾î ±¸¼º°ú ¼³Á¤À» Á¦¾ÈÇÑ´Ù. »ó±âÇÑ ½Ãµµ¸¦ ÅëÇØ Á¦¾ÈÇÏ´Â SmartX AI ÄÄÇ»Æà Ŭ·¯½ºÅÍ¿Í ½ÇÁ¦ ICT ÀÀ¿ë ¼­ºñ½ºµé°úÀÇ ¿¬°è¸¦ ÅëÇØ È¿°úÀûÀÎ ¼­ºñ½º Áö´ÉÈ­ÀÇ °¡´É¼ºÀ» »ìÆ캻´Ù.
¿µ¹®³»¿ë
(English Abstract)
With the increasing adoption of Intelligent ICT Application Services on Micro Service Architecture (MSA), there has been a rising need for an effective and flexible AI Computing Resource. In this paper, we propose an alternative concept of a Cloud-native-based SmartX AI Computing cluster that can effectively cope with the rising AI demands. Container virtualization & container orchestration, which are key components of cloud-native computing, enable the dynamical and flexible deployment and operation of distributed BigData/ML workloads. The proposed cluster components are based on AI-driven high-performance hardware and open-source software. In this paper, we design the hardware structure and the software stack required to adopt the proposed cluster concept into actual real-world clusters. We also implement a small cluster to examine its functionality and possible uses. The paper further expands the topic by providing the results of the experiments running real-distributed BigData/Machine-learning training workloads as well as identifying bottleneck points caused by storage and networking issues. The bottleneck points are used to propose the optimized software stack and configuration required to overcome such issues. The potential of effective intelligent service is presented in the proposed SmartX AI computing cluster when linked to actual ICT application services. Keywords: cloud-native computing, distributed AI computing, bigdata computing, machine learning,
Å°¿öµå(Keyword) Ŭ¶ó¿ìµå-³×ÀÌƼºê ÄÄÇ»Æà  ºÐ»ê ÀΰøÁö´É ÄÄÇ»Æà  ºòµ¥ÀÌÅÍ ÄÄÇ»Æà  ¸Ó½Å·¯´×   ÄÁÅ×ÀÌ³Ê ±â¹Ý Ŭ·¯½ºÅÍ   ¿ÀǼҽº ¼ÒÇÁÆ®¿þ¾î   cloud-native computing   distributed AI computing   bigdata computing   machine learning   container-enabled cluster   open-source software collaboration  
ÆÄÀÏ÷ºÎ PDF ´Ù¿î·Îµå