• Journal of Platform Technology
• /
• v.11 no.4
• /
• pp.71-79
• /
• 2023

In this paper, we analyze the characteristics of machine learning workloads and, based on them, propose a distributed in-memory caching technique to improve the performance of machine learning workloads. The core of machine learning workload is model training, and model training is a computationally intensive task. Performing machine learning workloads in a Kubernetes-based cloud environment in which the computing framework and storage are separated can effectively allocate resources, but delays can occur because IO must be performed through network communication. In this paper, we propose a distributed in-memory caching technique to improve the performance of machine learning workloads performed in such an environment. In particular, we propose a new method of precaching data required for machine learning workloads into the distributed in-memory cache by considering Kubflow pipelines, a Kubernetes-based machine learning pipeline management tool.

• Journal of Korea Society of Industrial Information Systems
• /
• v.29 no.5
• /
• pp.11-19
• /
• 2024

As task offloading from user devices transitions from the cloud to the edge, the demand for efficient resource management techniques has emerged. While numerous studies have employed reinforcement learning to address this challenge, many fail to adequately consider the overhead associated with real-world offloading tasks. This paper proposes a reinforcement learning-based distributed offloading policy generation method that incorporates task overhead. A simulation environment is constructed to validate the proposed approach. Experimental results demonstrate that the proposed method reduces edge queueing time, achieving up to 46.3% performance improvement over existing approaches.

• Journal of KIISE
• /
• v.44 no.10
• /
• pp.1112-1123
• /
• 2017

As content providers further offload content-centric services to the cloud, data retrieval over the cloud typically results in many redundant items because there is a prevalent near-duplication of content on the Internet. Simply fetching all data from the cloud severely degrades efficiency in terms of resource utilization and bandwidth, and data can be encrypted by multiple content providers under different keys to preserve privacy. Thus, locating near-duplicate data in a privacy-preserving way is highly dependent on the ability to deduplicate redundant search results and returns best matches without decrypting data. To this end, we propose an efficient near-duplicate detection scheme for encrypted data in the cloud. Our scheme has the following benefits. First, a single query is enough to locate near-duplicate data even if they are encrypted under different keys of multiple content providers. Second, storage, computation and communication costs are alleviated compared to existing schemes, while achieving the same level of search accuracy. Third, scalability is significantly improved as a result of a novel and efficient two-round detection to locate near-duplicate candidates over large quantities of data in the cloud. An experimental analysis with real-world data demonstrates the applicability of the proposed scheme to a practical cloud system. Last, the proposed scheme is an average of 70.6% faster than an existing scheme.

• Journal of Internet Computing and Services
• /
• v.19 no.1
• /
• pp.37-47
• /
• 2018

The number of capabilities of Internet of Things (IoT) devices will exponentially grow over the next years. These devices may generate a vast amount of time-constrained data. In the context of IoT, data management should act as a layer between the objects and devices generating the data and the applications accessing the data for analysis purposes and services. In addition, most of IoT services will be content-centric rather than host centric to increase the data availability and the efficiency of data delivery. IoT will enable all the communication devices to be interconnected and make the data generated by or associated with devices or objects globally accessible. Also, fog computing keeps data and computation close to end users at the edge of network, and thus provides a new breed of applications and services to end users with low latency, high bandwidth, and geographically distributed. In this paper, we propose Edge-Fog cloud-based Hierarchical Data Delivery ($EFcHD^2$) method that effectively and reliably delivers IoT data to associated with IoT applications with ensuring time sensitivity. The proposed $EFcHD^2$ method stands on basis of fully decentralized hybrid of Edge and Fog compute cloud model, Edge-Fog cloud, and uses information-centric networking and bloom filters. In addition, it stores the replica of IoT data or the pre-processed feature data by edge node in the appropriate locations of Edge-Fog cloud considering the characteristic of IoT data: locality, size, time sensitivity and popularity. Then, the performance of $EFcHD^2$ method is evaluated through an analytical model, and is compared to fog server-based and Content-Centric Networking (CCN)-based data delivery methods.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.34 no.2
• /
• pp.263-279
• /
• 2024

With the increasing trend of Cloud migration, security threats in the Cloud computing environment have also experienced a significant increase. Consequently, the importance of efficient incident investigation through log data analysis is being emphasized. In Cloud environments, the diversity of services and ease of resource creation generate a large volume of log data. Difficulties remain in determining which events to investigate when an incident occurs, and examining all the extensive log data requires considerable time and effort. Therefore, a systematic approach for efficient data investigation is necessary. CloudTrail, the Amazon Web Services(AWS) logging service, collects logs of all API call events occurring in an account. However, CloudTrail lacks insights into which logs to analyze in the event of an incident. This paper proposes an automated analysis framework that integrates Cloud Matrix and event information for efficient incident investigation. The framework enables simultaneous examination of user behavior log events, event frequency, and attack information. We believe the proposed framework contributes to Cloud incident investigations by efficiently identifying critical events based on the ATT&CK Framework.

• Journal of Internet Computing and Services
• /
• v.14 no.6
• /
• pp.71-84
• /
• 2013

Log data, which record the multitude of information created when operating computer systems, are utilized in many processes, from carrying out computer system inspection and process optimization to providing customized user optimization. In this paper, we propose a MongoDB-based unstructured log processing system in a cloud environment for processing the massive amount of log data of banks. Most of the log data generated during banking operations come from handling a client's business. Therefore, in order to gather, store, categorize, and analyze the log data generated while processing the client's business, a separate log data processing system needs to be established. However, the realization of flexible storage expansion functions for processing a massive amount of unstructured log data and executing a considerable number of functions to categorize and analyze the stored unstructured log data is difficult in existing computer environments. Thus, in this study, we use cloud computing technology to realize a cloud-based log data processing system for processing unstructured log data that are difficult to process using the existing computing infrastructure's analysis tools and management system. The proposed system uses the IaaS (Infrastructure as a Service) cloud environment to provide a flexible expansion of computing resources and includes the ability to flexibly expand resources such as storage space and memory under conditions such as extended storage or rapid increase in log data. Moreover, to overcome the processing limits of the existing analysis tool when a real-time analysis of the aggregated unstructured log data is required, the proposed system includes a Hadoop-based analysis module for quick and reliable parallel-distributed processing of the massive amount of log data. Furthermore, because the HDFS (Hadoop Distributed File System) stores data by generating copies of the block units of the aggregated log data, the proposed system offers automatic restore functions for the system to continually operate after it recovers from a malfunction. Finally, by establishing a distributed database using the NoSQL-based Mongo DB, the proposed system provides methods of effectively processing unstructured log data. Relational databases such as the MySQL databases have complex schemas that are inappropriate for processing unstructured log data. Further, strict schemas like those of relational databases cannot expand nodes in the case wherein the stored data are distributed to various nodes when the amount of data rapidly increases. NoSQL does not provide the complex computations that relational databases may provide but can easily expand the database through node dispersion when the amount of data increases rapidly; it is a non-relational database with an appropriate structure for processing unstructured data. The data models of the NoSQL are usually classified as Key-Value, column-oriented, and document-oriented types. Of these, the representative document-oriented data model, MongoDB, which has a free schema structure, is used in the proposed system. MongoDB is introduced to the proposed system because it makes it easy to process unstructured log data through a flexible schema structure, facilitates flexible node expansion when the amount of data is rapidly increasing, and provides an Auto-Sharding function that automatically expands storage. The proposed system is composed of a log collector module, a log graph generator module, a MongoDB module, a Hadoop-based analysis module, and a MySQL module. When the log data generated over the entire client business process of each bank are sent to the cloud server, the log collector module collects and classifies data according to the type of log data and distributes it to the MongoDB module and the MySQL module. The log graph generator module generates the results of the log analysis of the MongoDB module, Hadoop-based analysis module, and the MySQL module per analysis time and type of the aggregated log data, and provides them to the user through a web interface. Log data that require a real-time log data analysis are stored in the MySQL module and provided real-time by the log graph generator module. The aggregated log data per unit time are stored in the MongoDB module and plotted in a graph according to the user's various analysis conditions. The aggregated log data in the MongoDB module are parallel-distributed and processed by the Hadoop-based analysis module. A comparative evaluation is carried out against a log data processing system that uses only MySQL for inserting log data and estimating query performance; this evaluation proves the proposed system's superiority. Moreover, an optimal chunk size is confirmed through the log data insert performance evaluation of MongoDB for various chunk sizes.

• Journal of Intelligence and Information Systems
• /
• v.18 no.3
• /
• pp.35-51
• /
• 2012

Cloud computing is internet-based computing where computing resources are offered over the Internet as scalable and on-demand services. In particular, in case a number of various cloud services emerge in accordance with development of internet and mobile technology, to select and provide services with which service users satisfy is one of the important issues. Most of previous works show the limitation in the degree of user satisfaction because they are based on so called concept similarity in relation to user requirements or are lack of versatility of user preferences. This paper presents cloud service selection reasoning which can be applied to the general cloud service environments including a variety of computing resource services, not limited to web services. In relation to the service environments, there are two kinds of services: atomic service and composite service. An atomic service consists of service attributes which represent the characteristics of service such as functionality, performance, or specification. A composite service can be created by composition of atomic services and other composite services. Therefore, a composite service inherits attributes of component services. On the other hand, the main participants in providing with cloud services are service users, service suppliers, and service operators. Service suppliers can register services autonomously or in accordance with the strategic collaboration with service operators. Service users submit request queries including service name and requirements to the service management system. The service management system consists of a query processor for processing user queries, a registration manager for service registration, and a selection engine for service selection reasoning. In order to enhance the degree of user satisfaction, our reasoning stands on basis of the degree of conformance to user requirements of service attributes in terms of functionality, performance, and specification of service attributes, instead of concept similarity as in ontology-based reasoning. For this we introduce so called a service attribute graph (SAG) which is generated by considering the inclusion relationship among instances of a service attribute from several perspectives like functionality, performance, and specification. Hence, SAG is a directed graph which shows the inclusion relationships among attribute instances. Since the degree of conformance is very close to the inclusion relationship, we can say the acceptability of services depends on the closeness of inclusion relationship among corresponding attribute instances. That is, the high closeness implies the high acceptability because the degree of closeness reflects the degree of conformance among attributes instances. The degree of closeness is proportional to the path length between two vertex in SAG. The shorter path length means more close inclusion relationship than longer path length, which implies the higher degree of conformance. In addition to acceptability, in this paper, other user preferences such as priority for attributes and mandatary options are reflected for the variety of user requirements. Furthermore, to consider various types of attribute like character, number, and boolean also helps to support the variety of user requirements. Finally, according to service value to price cloud services are rated and recommended to users. One of the significances of this paper is the first try to present a graph-based selection reasoning unlike other works, while considering various user preferences in relation with service attributes.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.13 no.6
• /
• pp.2707-2712
• /
• 2012

Desktop-based computing environment is changed into mobile computing using smart phone and cloud computing providing common behavior and big data by network. Because of this transformation software development and operating environment is changed into heterogeneous distributed environment. As a result, dynamic service composition or changement is required. However, there is few research of techniques supporting service composition or changement dynamically in this situation. This paper suggests a technique for customizing services dynamically of mobile applications based on android platform. Especially we propose a customization technique of service by applying ontology technique to improve sharing and reuse of service. We applied proposed technique into meeting notification system, and obtain that it can be customized into various services such as email, sms, twitter service, and so on.

• KIISE Transactions on Computing Practices
• /
• v.21 no.1
• /
• pp.1-12
• /
• 2015

To realize futuristic services with agility, the role of the experimental facility (i.e., testbed) based on integrated resources has become important, so that developers can flexibly utilize the dynamic provisioning power of software-defined networking and cloud computing. Following this trend, an OpenFlow-based SDN testbed environment, denoted as OF@TEIN, connects multiple sites with unique SmartX Racks (i.e., virtualization-enabled converged resources). In this paper, in order to automate the multi-point L2 (i.e., Ethernet) inter-connection of OpenFlow islands, we introduce an automated tool to configure the required Network Virtualization using Generic Routing Encapsulation (NVGRE) tunneling. With the proposed automation tool, the operators can efficiently and quickly manage network inter-connections among multiple OpenFlow sites, while letting developers to control their own traffic flows for service realization experiments.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.40 no.1
• /
• pp.61-69
• /
• 2015

Through Virtual Machine technology(VM), VMs can be packed into much fewer number of physical servers than that of VMs. Since even an idle physical server wastes more than 60% of max power consumption, it has been considered as one of energy saving technologies to minimize the number of physical servers by using the knapsack problem solution based on the computing resources. However, this paper shows that this tightly packed consolidation may not achieve the efficient energy saving. Instead, a service pattern-based VM consolidation algorithm is proposed. The proposed algorithm takes the service time of each VM into account, and consolidates VMs to physical servers in the way to minimize energy consumption. The comprehensive simulation results show that the proposed algorithm gains more than 30% power saving.