Abstract: Provided are a method and apparatus for obtaining a minimum cost vector for making a skyline object in a multi-dimensional space. The method includes calculating respective vector values having a query point and respective moving points to which the query point is moved as both end points in a multi-dimensional space having a plurality of coordinate axes, and selecting a vector value whose moving point is included in a skyline and has the minimum distance value from the query point as the minimum vector value from among the vector values.

Abstract: Disclosed is a dynamic load balancing system. The dynamic load balancing system includes a resource management master managing bare servers that do not execute services and having a hierarchical structure and a service master dynamically allocating the bare servers to a load balancing server or a service execution server or dynamically releasing the pre-allocated load balancing server or service execution server by the bare servers, in consideration of monitoring information on a state or performance of a server and service requirements to be provided.

Abstract: Disclosed is a method for moving a personal virtual machine among clouds in a cloud computing environment, including: requesting metadata for driving a virtual machine allocated to a user and an updated file in a first cloud; generating a virtual machine image in a second cloud based on the metadata and the updated file; and reproducing the virtual machine by using the generated virtual machine image.

Abstract: Disclosed herein is a method for deduplication of virtual machine images, including: generating a plurality of chunks by dividing the virtual machine images into predetermined units; determining chunks corresponding to identifiers previously stored in a repository as chunks previously stored in the repository using identifiers for each of the plurality of chunks; storing chunks that have not been stored in the repository among the plurality of chunks in the repository using the chunks previously stored in the repository; and storing inner position information on each of the plurality of chunks for the virtual machine images in image specifications corresponding to the virtual machine images.

Abstract: Provided are a large-scale cluster monitoring system and a method for automatically building/restoring the same, which can automatically build a large-scale monitoring system and can automatically build a monitoring environment when a failure occurs in nodes. The large-scale cluster monitoring system includes a CM server, a BD server, GM nodes, NA nodes, and a DB agent. The CM server manages nodes in a large-scale cluster system. The DB server stores monitoring information that is state information of nodes in groups. The GM nodes respectively collect the monitoring information that is the state information of the nodes in the corresponding groups to store the collected monitoring information in the DB server. The NA nodes access the CM server to obtain GM node information and respectively collect the state information of the nodes in the corresponding groups to transfer the collected state information to the corresponding GM nodes.

Abstract: Provided is an availability prediction method for a high availability. The method includes calculating a basic survival probability that the other node survives until a failure on one node of two nodes constituting a cluster is fixed, and determining an optimal number of nodes meeting a preset reference availability probability by calculating an availability probability for a predetermined range of the number of nodes on the basis of the basic survival probability. The method determines the number of nodes in the high availability cluster so as to match a reference availability probability, and is able to accomplish an optimal configuration of a cluster by calculating the availability probabilities for combinations between active node and passive nodes and between head nodes and switches.

Abstract: Provided are a method and apparatus for obtaining a minimum cost vector for making a skyline object in a multi-dimensional space. The method includes calculating respective vector values having a query point and respective moving points to which the query point is moved as both end points in a multi-dimensional space having a plurality of coordinate axes, and selecting a vector value whose moving point is included in a skyline and has the minimum distance value from the query point as the minimum vector value from among the vector values.

Abstract: A method of data placement and control in a block-divided distributed disk array is provided. At first, data to store is divided into logical blocks, and each of the divided logical blocks is further divided into a plurality of data blocks. Then, a parity block is created through performing an XOR operation on the data blocks of each logical block in a unit of a strip, and the parity block is arranged at the first block of a strip unit. Then, parity blocks are arranged at a next block of a previous logical block's the last data block through a left/right shifting for uniformly arranging the parity blocks through out all disks in order to fill an empty block between the logical blocks.

Abstract: Provided are a cluster system, which makes general nodes appear as if they provide seamless services without failure when seen from the outside, and a method for operating the cluster system. The cluster system for operating individual nodes in a distributed management manner includes a board server having a task board registered with a task list, an agent server for managing the task board, and a plurality of general server nodes for performing a corresponding task on the basis of the task list, among which a failed general server node is replaced with another normal general server node.

Abstract: Provided are a large-scale cluster monitoring system and a method for automatically building/restoring the same, which can automatically build a large-scale monitoring system and can automatically build a monitoring environment when a failure occurs in nodes. The large-scale cluster monitoring system includes a CM server, a BD server, GM nodes, NA nodes, and a DB agent. The CM server manages nodes in a large-scale cluster system. The DB server stores monitoring information that is state information of nodes in groups. The GM nodes respectively collect the monitoring information that is the state information of the nodes in the corresponding groups to store the collected monitoring information in the DB server. The NA nodes access the CM server to obtain GM node information and respectively collect the state information of the nodes in the corresponding groups to transfer the collected state information to the corresponding GM nodes.

Abstract: Provided is an availability prediction method for a high availability. The method includes calculating a basic survival probability that the other node survives until a failure on one node of two nodes constituting a cluster is fixed, and determining an optimal number of nodes meeting a preset reference availability probability by calculating an availability probability for a predetermined range of the number of nodes on the basis of the basic survival probability. The method determines the number of nodes in the high availability cluster so as to match a reference availability probability, and is able to accomplish an optimal configuration of a cluster by calculating the availability probabilities for combinations between active node and passive nodes and between head nodes and switches.

Abstract: A data acceleration apparatus for Internet small computer system interface (iSCSI) and an iSCSI storage system using the same are provided. A data acceleration apparatus comprises an I/O processor, a memory, a TOE controller, and an IPsec controller. The I/O processor is connected to an iSCSI storage system through a PCI bus and controls a PCI bridge and the memory. The memory stores data according to the control by the I/O processor, serves as a buffer for processing a TCP/IP and an IPsec protocol, and provides a data storage space. The TOE controller receives data to be read or written through the I/O processor, offloads a stack of the TCP/IP, and processes the TCP/IP. The IPsec controller processes the IPsec protocol for the inputted and outputted data.

Abstract: A method of data placement and control in a block-divided distributed disk array is provided. At first, data to store is divided into logical blocks, and each of the divided logical blocks is further divided into a plurality of data blocks. Then, a parity block is created through performing an XOR operation on the data blocks of each logical block in a unit of a strip, and the parity block is arranged at the first block of a strip unit. Then, parity blocks are arranged at a next block of a previous logical block's the last data block through a left/right shifting for uniformly arranging the parity blocks through out all disks in order to fill an empty block between the logical blocks.