취업 성공사례 - 뜻을 세워 미래로, 꿈을 향해 세계로! 호남대학교!

윤천균,Yun, Cheon-Gyun 한국대학교육협의회 2009 大學敎育 Vol.161 No.-

웹 클러스터 시스템의 실시간 서버 상태를 기반으로 한 부하분산 방안

윤천균,Youn, Chun-Kyun 한국정보처리학회 2005 정보처리학회논문지 A Vol.12 No.5

I studied about existent load distribution algorithms and the WLC(Weighted Least Connection) algerian that is using much at present to distribute the connection request of users to real servers efficiently in web cluster system. The efficiency of web cluster system is fallen by load imbalance between servers, because there is problem In inaccurate load status measuring of servers and measuring timing at these load distribution algorithms. In this paper, I suggest an algorithm that distributes load base on various load state of servers by real time using broadcasting RPC(Remote Procedure Call) when a user requests connection, and implement a prototype and experiment its performance. The experiment result shows that load imbalance phenomenon between reai sowers was improved greatly than existing method, and performance of web cluster system was improved by efficiency that response time is shortened. 웹 클러스터 시스템에서 사용자들의 접속 요구를 효율적으로 real 서버에 분산시키기 위하여 기존의 부하분산 알고리즘들과 현재 많이 사용중인 WLC(Weighted Least Connection) 알고리즘에 대하여 연구하였다. 이 부하분산 알고리즘들은 서버들의 부정확한 부하상태 파악과 파악 시점에 문제가 있어 서버들 간에 부하 불균형이 발생함으로써 효율성이 저하된다. 본 연구에서는 사용자의 접속 요청 시 broadcasting RPC(Remote Procedure Call)를 이용하여 서버들의 다양한 부하상태를 실시간으로 파악하여 부하를 분산하는 알고리즘을 제안하고, prototype을 구현하여 성능을 실험하였다. 실험 결과, real 서버 간 부하 불균형 현상이 기존 방법에 비하여 크게 향상되었고, 응답시간이 단축되는 효과로 웹 클러스터 시스템의 성능이 향상되었다.

저부하의 네트워크 트래픽을 갖는 효율적인 SNMP 설계

윤천균,이찬민,탁동길,정일용 한국통신학회 2003 韓國通信學會論文誌 Vol.28 No.4B

부하분산 클러스터 시스템에서의 효율적인 가중치 적용 알고리즘 연구

윤천균 호남대학교 정보통신연구소 2004 정보통신연구 Vol.14 No.-

The importance of the Internet in our daily life and it's fast growth gave consequence to its importance. So the traffic load of Internet is increasing dramatically and it causes system overload and a bottle neck of network. Therefore, in order to solve these problems, the demand for clustering system with high extension and high availability server. There is a limitation in single server solution which enables server to make high performance server. Reformations could be complicated or it will waste of the original system. And if demanding from customers keep increasing more than the system going to be overloaded again and it'll need another reformation of the server. But The higher the goal of the, reformations gets, the more cost is expected. So the way of solving these problem is to research the web clustering system actively. This paper proposes a more efficient load distribution of the web clustering system based on RPC dynamic weight to improve the fixed weight method. We implemented a prototype system and tested the performance of both methods. The proposed clustering algorithm can distribute more efficiently the load of web servers with real time measuring technic of the real server's status. RPC library is using to measure status of real servers. We was proposed to take better response time rate of application layer to be sure efficient load distribution through computed weighted value based on free memory and network capacity of the real server. This result of the experimentation, showed that the application of RPC dynamic weighted value improved memory availability as about 3.5 times as that of to WRR fixed weight method and as about 10 times as WLC. And made more efficient of memory availability through equalize free memory of the real server. Response time in the aspect of by applying dynamic weight method to a little load of server based on RPC real time measuring technic of the real server's status, it become possible to get rapid response time with real time measuring technic of the real server's status and confirm 9.5% performance development as the same-time clients increase. RPC service got the status information with rarely affecting another service because it put the slight load on real server.

네트워크 관리 프로토콜(SNMP) 기능 확장을 통한 네트워크 부하 감소 연구

윤천균,정일용 조선대학교 전자정보통신연구소 2002 電子情報通信硏究所論文誌 Vol.5 No.2

A SNMP manager requests a MIB value to an agent for a MIB object, and then the agent responds to the manager in the SNMP which is used widely in the world. By this reason, so much traffic is produced between the manger and the agent of SNMP. Specially, the recursive sending-receiving of SNMP information for the trend analysis function of network management is very inefficient. In this paper, an effective method is proposed to minimize the unnecessary sending -receiving information between Manager and Agent for collecting the trend management information. The trend management information(TMI) PDUs are added to manage some time functions in the existent SNMP. The proposed SNMP method is implemented to keep compatibility with the existent SNMP function. The test results of the improved SNMP show that it has compatibility with the existent SNMP and the network traffic load was largely decreased.

경향분석 정보의 효율적인 처리를 통한 네트워크 트래픽 감소 방안에 대한 연구

윤천균 한국디지털정책학회 2012 디지털융복합연구 Vol.10 No.1

시작시각 기반의 효율적인 WFQ 패킷 스케줄러 연구

윤천균 호남대학교 정보통신연구소 2005 정보통신연구 Vol.15 No.-

The times tamp based fair packet scheduling algorithms can be divided into the following two schemes with regard to the reference time used in calculating the time stamp of arriving packet the Finish-Time (FT) and Start-Time (ST) schemes. While in the FT scheme the expected transmission finish time of the packet is used as its time stamp, its expected transmission start time is used in the ST one. The FT scheme has the property of an inversely rate-proportional latency but the latency in the ST one is relying on the maximum allowable number of flows in scheduler. In a consequence, the FT scheme is suitable to support multiple delay bounds but there may be some bandwidth loss due to excess reservation. Meanwhile the ST one has an advantage in reducing the bandwidth loss but it is hard to support multiple delay bounds because of its latency property relying on the number of flows. This paper is devoted to propose a ST based scheduler to effectively support multiple delay bounds and analyze its performance comparing to the FT based scheduler. The comparison results show that the proposed scheduler gives better utilization by up to 50%.

IPC 개선을 통한 라우터의 성능향상 방안 연구

윤천균 호남대학교 2004 호남대학교 학술논문집 Vol.25 No.2

Network performance has been decreased because of the bottle-neck of router system by explosive increase of Internet traffic according to users increase and fast increase of high speed multimedia service. we suggested R-IPC method which improve existent Ethernet IPC that transmit data through transmission layer and Internet layer to solve this problem. The proposing R-IPC which is applied to a prototype system shows performance improvement of network because that processing is faster than existing communication method, When exchange information which is connected with routing table between routers in distributed network structure such as figure 1. At the same time, R-IPC method shows packet processing performance improvement effect of average 3.26% for OAM function which is processed at layer 1 and 2 in Router system inside. Conclusively, the performance of Router is improved by applying R-IPC method than existent Router data processing method.

웹 클러스터 시스템의 실시간 서버 상태를 기반으로 한 부하분산 방안

윤천균 한국정보처리학회 2005 정보처리학회논문지. 컴퓨터 및 통신시스템 Vol.12 No.5

웹 클러스터 시스템에서 사용자들의 접속 요구를 효율적으로 real 서버에 분산시키기 위하여 기존의 부하분산 알고리즘들과 현재 많이 사용 중인 WLC(Weighted Least Connection) 알고리즘에 대하여 연구하였다. 이 부하분산 알고리즘들은 서버들의 부정확한 부하상태 파악과 파악 시점에 문제가 있어 서버들 간에 부하 불균형이 발생함으로써 효율성이 저하된다. 본 연구에서는 사용자의 접속 요청 시 broadcasting RPC(Remote Procedure Call)를 이용하여 서버들의 다양한 부하상태를 실시간으로 파악하여 부하를 분산하는 알고리즘을 제안하고, prototype을 구현하여 성능을 실험하였다. 실험 결과, real 서버 간 부하 불균형 현상이 기존 방법에 비하여 크게 향상되었고, 응답시간이 단축되는 효과로 웹 클러스터 시스템의 성능이 향상되었다. I studied about existent load distribution algorithms and the WLC(Weighted Least Connection) algorithm that is using much at present to distribute the connection request of users to real servers efficiently in web cluster system. The efficiency of web cluster system is fallen by load imbalance between servers, because there is problem in inaccurate load status measuring of servers and measuring timing at these load distribution algorithms. In this paper, I suggest an algorithm that distributes load base on various load state of servers by real time using broadcasting RPC(Remote Procedure Call) when a user requests connection, and implement a prototype and experiment its performance. The experiment result shows that load imbalance phenomenon between real servers was improved greatly than existing method, and performance of web cluster system was improved by efficiency that response time is shortened.

SNMP PDU의 시간변수 추가를 통한 네트워크 모니터링 성능 향상에 관한 연구

윤천균(Chun-Kyun Youn),정일용(Il-Yong Chung) 한국멀티미디어학회 2003 멀티미디어학회논문지 Vol.6 No.7