High Availability Support Research Articles

Fault tolerance strategy to increase Moodle service reliability

Moodle is one of online Learning Management System that can be downloaded and installed on our server for free. Moodle is suitable solution for universities to organize online learning. Changing all learning activities from offline to online during a pandemic situation, especially for universities, requires Moodle services to be reliably available. The more students it means the more traffic should be handled by Moodle server. Moodle consists of system, user’s data and a database. To increase Moodle reliability, we can rely on these components. We can use virtualization technology to simplify migration system and recovery process. Distributed file system technology is a best solution to store Moodle user data because it can be replicated into several nodes and support for high availability. Finally, the last component of Moodle is a database, we can replicate the database into several machine to increase availability. At the top of every Moodle component, we need cluster control to make sure all of Moodle components can be accessed. The results show that with the right fault tolerance strategy, all requests can be responded to 100% without error. In addition, this strategy can also increase the response time 12 times faster when compared to a single server.

APPLICATION OF COMPUTATIONAL MODELS TO DETECTION OF OUTAGE REASONS IN CLOUD INFRASTRUCTURES

Goal. Monitoring of technical status is one of the most important tasks in modern cloud infrastructures. One of the most popular solution is providing list of currently known outages to operator; that might be inconvenient in case of necessity to detect reason of multiple dependant outages. Thus it is necessary to inform operators how outages are related to each other. Materials and methods. The paper proposes monitoring of cloud infrastructure based upon comparison between their functioning metrics and golden model. It also proposes usage of Tyugu’s computational modes for defining this golden model and also for defining visual representation of cloud infrastructure. Results. Developed methods for describing technical condition of cloud application using computational models, detection of complex outage reasons, visualization of parts of the application relevant to outage. Conclusion. It is demonstrated that the approach has some advantages, namely, allows automatically detect reasons of outages and visualize parts of applications, relevant to these outages. At the same time there are some difficulties adopting this approach in systems with complex high availability support strategies.

NoMISHAP: A Novel Middleware Support for High Availability in Multicloud PaaS

The availability of multiple Platform as a Service (PaaS) environments, characterized by different actors, supported service models, and available components ease the engineering of new applications and facilitate their fast composition and elastic delivery. However, as demonstrated by the several outages that affected all main cloud providers along the last decade, when these applications come to a production stage there are still several open issues, especially for applications that must meet high availability requirements. This paper proposes NoMISHAP, a Novel Middleware Support for High Availability in multicloud PaaS. NoMISHAP provides the needed middleware support to develop of a new class of highly available services over multiple PaaS environments: it includes service component adapters to ease service migration between different PaaS and supports all needed monitoring, brokerage, and binding/deployment operations to enable transparent transitions between PaaS environments as failures occur at runtime. We deeply assessed NoMISHAP in a realistic scenario over Heroku, Bluemix, and Openshift PaaS platforms. Collected results show both the effectiveness of our solution and quantify the overhead, with the goal of demonstrating the suitability of middleware-based multicloud PaaS solutions to grant high availability.

Evaluation of redundant data storage in clusters based on multi-multicast and local storage

Cluster platforms have an important role in high performance computing (HPC). They execute cloud computing, data-intensive computing and data center applications, which are supported on distributed file systems. The implementation of data redundancy in these file systems provides a support for high availability and error tolerance. This work proposes an implementation of redundant data storage based on the storage included in the cluster nodes, instead of more expensive approaches with a dedicated storage and network, and on multi-multicast transfers, instead on unicast transfers, to perform the multiple simultaneous data diffusion required for implementing redundant data storage. The proposal applies a recently proposed congestion control scheme that adjusts the sender injection rate, taking into account control information from the receiver nodes and the storage technology available on the cluster nodes. The implementation takes full advantage of the switch diffusion hardware and of the IGMP snooping capability of current switches, which allows to multicast a packet just to the output links with receivers joined to a multicast group. It is made at the user level directly on the UDP interface. Evaluation tests with multiple simultaneous storage accesses were performed in a CentOS cluster. Test results show a more efficient use of the cluster storage. The global bandwidth improves by using hardware related to the storage (network and storage devices) more efficiently.

Modern main-memory database systems

This tutorial provides an overview of recent developments in main-memory database systems. With growing memory sizes and memory prices dropping by a factor of 10 every 5 years, data having a "primary home" in memory is now a reality. Main-memory databases eschew many of the traditional architectural tenets of relational database systems that optimized for disk-resident data. Innovative approaches to fundamental issues such as concurrency control and query processing are required to unleash the full performance potential of main-memory databases. The tutorial is focused around design issues and architectural choices that must be made when building a high performance database system optimized for main-memory: data storage and indexing, concurrency control, durability and recovery techniques, query processing and compilation, support for high availability, and ability to support hybrid transactional and analytics workloads. This will be illustrated by example solutions drawn from four state-of-the-art systems: H-Store/VoltDB, Hekaton, HyPeR, and SAP HANA. The tutorial will also cover current and future research trends.

High Available VoIP Server Failover Mechanism in Wide Area Network

The development of telecommunication technology specified the Internet Protocol (IP) based technology for the next generation network. Voice over Internet Protocol (VoIP) has been introduced to overcome future telephony demand. However, these rapid changes encountered some issues, and the most critical is how to provide the services availability and reliability equally to circuit based telephony. Virtualization is widely used not only for hardware efficiency and maintenance, but also for High Availability support. Virtualized environment provides the ability among servers to migrate or replicate into another machine, even when they are running their services, which is known as Live Migration. In this paper, the ability of virtualization failover mechanism over WAN and voice quality of VoIP service when running on the High Availability system in virtualized environment is studied and examined. The objective analysis by using quality of services (QoS) attributes is conducted as well as the subjective analysis using Mean Opinion Score (MOS). The work utilizes Xen® Hypervisor with modified Remus extensions to provide the High Availability environment with GRE tunnelling and newtwork virtualization. Remus approach using checkpoint based is deployed to copy the primary server to the backup server. A range of 40ms – 900ms has been applied as time interval of checkpoint. The results show that the the failover downtime is 1.4 s and mean jitter is 9,98 ms, packet loss 3,12% and MOS 3.61 for Remus 400ms checkpoint. MOS with different checkpoint time interval is also presented.

BlueStar: A federation-based approach to building Internet-scale data centers

This paper describes the architecture and the implementation of BlueStar, which uses a federation-based approach to design and build Internet-scale data centers. Instead of aggregating all of the compute, storage, and network resources in a single data center, our approach uses the concept of a federation in which different-size data centers participate to provide the capacity needed to support Internet-scale applications. The main advantage of our approach is that due to the inherent geographic distribution of the data centers, applications can be easily provided with support for high availability, disaster recovery, and localized access (reduced latency). The approach also enables the service providers to realize benefits by deciding to locate the hosting of applications to take advantage of savings due to cost differentials between peak and off-peak hours, power rates, and cooling costs across data centers. Further, the approach enables individual data centers to maintain independence and operational control over their resources. We discuss the implementation of BlueStar using two sample composite applications: BlueMaps, a server-side mashup application (i.e., an application that provides data or functionality from more than one location), and Lotus® Connections, a collaboration application.