Kyrre M. Begnum

Principle Components and Importance Ranking of Distributed Anomalies

Correlations between locally averaged host observations, at different times and places, hint at information about the associations between the hosts in a network. These smoothed, pseudo-continuous time-series imply relationships with entities in the wider environment. For anomaly detection, mining this information might provide a valuable source of observational experience for determining comparative anomalies or rejecting false anomalies. The difficulties with distributed analysis lie in collating the distributed data and in comparing observables on different hosts, in different frames of reference. In the present work, we examine two methods (Principle Component Analysis and Eigenvector Centrality) that shed light on the usefulness of comparing data destined for different locations in a network.

IncludeOS: A Minimal, Resource Efficient Unikernel for Cloud Services

The emergence of cloud computing as a ubiquitous platform for elastically scaling services has generated need and opportunity for new types of operating systems. A service that needs to be both elastic and resource efficient needs A) highly specialized components, and B) to run with minimal resource overhead. Classical general purpose operating systems designed for extensive hardware support are by design far from meeting these requirements. In this paper we present IncludeOS, a single tasking library operating system for cloud services, written from scratch in C++. Key features include: extremely small disk-and memory footprint, efficient asynchronous I/O, OS-library where only what your service needs gets included, and only one device driver by default (virtio). As a test case a bootable disk image consisting of a simple DNS server with OS included is shown to require only 158 kb of disk space and to require 5-20% less CPU-time, depending on hardware, compared to the same binary running on Linux.

Simplified cloud-oriented virtual machine management with MLN

System administrators are faced with the challenge of making their existing systems power-efficient and scalable. Although cloud computing is offered as a solution to this challenge by many, we argue that having multiple interfaces and cloud providers can result in more complexity than before. This paper addresses cloud computing from a user perspective. We show how complex scenarios, such as an on-demand render farm and scaling web-service, can be achieved utilizing clouds but at the same time keeping the same management interface as for local virtual machines. Further, we demonstrate that by enabling the virtual machine to have its policy locally instead of in the underlying framework, it can move between otherwise incompatible cloud providers and sites in order to achieve its goals more efficiently.

Cloud-Oriented Virtual Machine Management with MLN

System administrators are faced with the challenge of making their existing systems power-efficient and scalable. Although Cloud Computing is offered as a solution to this challenge by many, we argue that having multiple interfaces and cloud providers can result in more complexity than before. This paper addresses cloud computing from a user perspective. We show how complex scenarios, such as an on-demand render farm and scaling web-service, can be achieved utilizing clouds but at the same time keeping the same management interface as for local virtual machines. Further, we demonstrate that by enabling the virtual machine to have its policy locally instead of in the underlying framework, it can move between otherwise incompatible cloud providers and sites in order to achieve its goals more efficiently.

On the stability of adaptive Service Level Agreements

We consider some implications of non-linear feedback, due to policy combinatorics, on policy-based management of networked services. We pay special attention to the case where the monitoring of certain aspects of Service Level Agreements is used to alter future policy dynamically, according to a control feedback scheme. Using two simple models, we show that nonlinear policies are generally unstable to service provision, i.e. provide no reliable service levels (QoS). Hence we conclude that automated control by policy-rule combinatorics can damage quality of service goals.

Adaptive provisioning using virtual machines and autonomous role-based management

A modern server system must deliver a complex set of obligations towards its users. We study the concept of roles as autonomous encapsulations of a set of promises. Using visualization and configuration management, every defined role for a real-life computer system can be realized as a self managing agent that is aware of other, related roles within the same broader context. A prototype implementation that provides web hotel services for customers is presented and discussed

A scaled, immunological approach to anomaly countermeasures: combining pH with cfengine

We discuss the combination of two anomaly detection models, the Linux kernel module pH and cfengine, in order to create a multi-scaled approach to computer anomaly detection with automated response. By examining the time-average data from pH, we find the two systems to be conceptually complementary and to have compatible data models. Based on these findings, we build a simple prototype system and comment on how the same model could be extended to include other anomaly detection mechanisms.

On the usefulness of off-the-shelf computer peripherals for people with Parkinson's Disease

People who suffer from Parkinson’s Disease face many challenges using computers, and mice are particularly problematic input devices. This article describes usability tests of standard peripherals for use by people with Parkinson’s Disease in order to identify optimal combinations with respect to the needs of this user group. The results are used to determine their effect upon inertia, muscle stiffness, tremor, pain, strain and coordination and show that widely available equipment could significantly improve mouse pointer control for many users. The results reflect the diversity of challenges experienced by computer users with Parkinson’s Disease, and also illustrate how projector-based technology may improve computer interaction without risking strain injuries.

Towards the InfiniBand SR-IOV vSwitch Architecture

To meet the demands of the Exascale era and facilitate Big Data analytics in the cloud while maintaining flexibility, cloud providers will have to offer efficient virtualized High Performance Computing clusters in a pay-as-you-go model. As a consequence, high performance network interconnect solutions, like InfiniBand (IB), will be beneficial. Currently, the only way to provide IB connectivity on Virtual Machines (VMs) is by utilizing direct device assignment. At the same time to be scalable, Single-Root I/O Virtualization (SR-IOV) is used. However, the current SR-IOV model employed by IB adapters is a Shared Port implementation with limited flexibility, as it does not allow transparent virtualization and live-migration of VMs. In this paper, we explore an alternative SR-IOV model for IB, the virtual switch (vSwitch), and propose and analyze two vSwitch implementations with different scalability characteristics. Furthermore, as network reconfiguration time is critical to make live-migration a practical option, we accompany our proposed architecture with a scalable and topology agnostic dynamic reconfiguration method, implemented and tested using OpenSM. Our results show that we are able to significantly reduce the reconfiguration time as route recalculations are no longer needed, and in large IB subnets, for certain scenarios, the number of reconfiguration subnet management packets (SMPs) sent is reduced from several hundred thousand down to a single one.

Xen Virtualization and Multi-host Management Using MLN

Xen virtualization is a powerful tool for encapsulating services and providing seamless migration of tasks on hardware failure. This tutorial shows how to set up multiple Xen instances in a network using the MLN management tool. One of the main challenges in virtual machine administration on a large scale is the specification of complex and repeatable virtualized scenarios. Creating a singe new virtual machine, boot it and install an operating system is straight forward with many of todays tools. But what if you need to deploy 50 identical virtual machines across 25 servers and manage them as an atomic unit? How do you at a later point make consistent design adjustments such as migrating a subset of the virtual machines to a new server or adjusting memory levels? These issues are at the heart of this tutorial. MLN is a virtual machine management tool supporting both User-Mode Linux and Xen. It has been developed at the Oslo University College in conjunction with its research on system administration and resource management. MLN and its research has previously been presented at the Norwegian Unix User Group (NUUG) and the 20th USENIX Large Installation System Administration conference LISA. This tutorial is interesting for all who want to look beyond the typical one-vm-on-my-desktop scenario. Teachers interested in virtual student labs should also attend. We start with a short introduction to the Xen virtual machine technology and then proceed to MLN and its own configuration language. In the second part of the tutorial, we will talk about installation and configuration of MLN into a virtual infrastructure across several servers.