Daniel Schlitt

Proactive dynamic resource management in virtualized data centers

Dynamically reassigning virtual machines (VMs) to servers is a widely addressed idea to save energy in data centers. VMs are consolidated in times of low overall resource demand. Unused servers are switched off to save energy. Mainly two major challenges must be addressed to realize this approach. First, the resource demand of VMs expected in the future must be estimated to take care of delays caused by VM migrations and server startups. An upcoming resource shortage must have been resolved right before it actually occurs. Second, a scheduling algorithm is needed that, based on a current distribution of VMs to servers, can guarantee to find a sequence of operations that resolves any upcoming resource shortage right in time. Within this paper, we present a novel approach that addresses both of these challenges. In contrast to previous work, this approach can guarantee not to cause any resource shortages, if the actual resource demand of the VMs meets the expected one. We performed a simulation based evaluation with a set of VMs. The underlying resource demand time series were measured in a data center operated by a medium-sized IT service provider. A data center model was used to estimate the energy consumption. Overall energy savings of about 23% could be achieved compared to a static approach. Resource shortages occurred in less than 0.1% of time. They could be resolved by the approach in less then 20 minutes.

Power and cost aware distributed load management

The continuously rising energy demand of data centers has already reached the two-digit megawatt area by now. The rising energy costs force operators to search for effective methods for energy reduction. A popular, software based instrument is consolidation using virtualization, since this provides also high flexibility for changing business requirements. Dynamic load management can introduce an even stronger consolidation here. An enterprise maintaining more than just one data center can also apply a distributed data center comprehensive load management. This paper presents a vision minimizing the energy requirement or energy costs, using geographical specific characteristics of servers and data centers. The potential for savings of the introduced distributed load management can be up to 40%.

The Energy Demand of Data Centers

Data centers are the backbone of today’s information technologies. With increasing usage of cloud services and web applications, the need for remote computing and storage will only grow. However, one has to consider that increasing numbers of server and storage systems also mean increases in energy consumption. The power demand is caused not only by the IT hardware, but is also due to the required infrastructure such as power supply and climatization. Therefore, choosing the most appropriate components as well as architectural designs and configurations regarding energy demand, availability, and performance is important. This chapter depicts influencing factors and current trends for these design choices and provides examples.

Gain More from PUE: Assessing Data Center Infrastructure Power Adaptability

The power usage effectiveness (PUE) for data centers is used by operators as KPI to measure the absolute infrastructure power overhead. However, this only draws conclusions on static or average operation conditions during an usual annual time period. For analyzing the aspect of dynamics in the IT to infrastructure power relation, we propose two new metrics. First, the power variability (PVar). It simply indicates the relative rates and heights of power variations. Second, the infrastructure power adaptability (IPA). It relates the power variabilities and relative average deviations of IT and infrastructure power in order to represent the scalability and adaptability of the infrastructure to the IT demands. Both metrics use the same input data also needed for a continuous PUE calculation. Thus, the applicability in a data center running a PUE-metering can be ensured. In an evaluation, we applied the IPA on power traces of a container data center (in the following denoted as CDC) and compared the results with PUE scalability, a metric with the same scope. The comparison showed, that IPA covers more operating states and is therefore more robust and reliable than its counterpart.

Load dependent data center energy efficiency metric based on component models

Common data center energy efficiency metrics only work on a high abstraction level and require actually measured values. With these metrics, it is not possible to identify the sources of shortcomings in efficiency or to explore possible changes in configuration or architecture, respectively. In this paper, an alternative metric addressing these drawbacks is introduced. The metric makes use of pre-characterized load dependent component models and estimates efficiency for arbitrary input data. The results are objectively comparable between different data center configurations as well as between data center sites, and reasons for inefficiencies may be identified by extracting intermediate results.

The M2DC Project: Modular Microserver DataCentre

The Modular Microserver DataCentre (M2DC) project will investigate, develop and demonstrate a modular, highly-efficient, cost-optimized server architecture composed of heterogeneous microserver computing resources, being able to be tailored to meet requirements from various application domains such as image processing, cloud computing or HPC. M2DC will be built on three main pillars: a flexible server architecture that can be easily customised, maintained and updated, advanced management strategies and system efficiency enhancements (SEE), well-defined interfaces to surrounding software data centre ecosystem.

Data centres for IoT applications: The M2DC approach (Invited paper)

The Modular Microserver DataCentre (M2DC) project investigates, develops and demonstrates a modular, highly-efficient, cost-optimized server architecture composed of heterogeneous micro server computing resources, being able to be tailored to meet requirements from various application domains, including the Internet of Things. M2DC is built on three main pillars: a flexible server architecture that can be easily customised, maintained and updated; advanced management strategies and system efficiency enhancements (SEE); well-defined interfaces to surrounding software data centre ecosystem.

M2DC – Modular Microserver DataCentre with heterogeneous hardware

The Modular Microserver DataCentre (M2DC) project investigates, develops and demonstrates a modular, highly-efficient, cost-optimized server architecture composed of heterogeneous microserver computing resources. The resulting server architecture will be able to be tailored to meet requirements from a wide range of application domains. M2DC is built on three main pillars: a flexible server architecture that can be easily customised, maintained and updated; advanced management strategies and system efficiency enhancements (SEE); well-defined interfaces to the surrounding software data centre ecosystem. In this paper, we focus in particular on the thermal management strategies and on the initial benchmarking of the Aarch64 ARM architecture.

M2DC – A Novel Heterogeneous Hyperscale Microserver Platform

The Modular Microserver Datacentre (M2DC) project targets the development of a new class of energy-efficient TCO-optimized appliances with built-in efficiency and dependability enhancements. The appliances will be easy to integrate with a broad ecosystem of management software and fully software defined to enable optimization for a variety of future demanding applications in a cost-effective way. The highly flexible M2DC server platform will enable customization and smooth adaptation to various types of applications, while advanced management strategies and system efficiency enhancements (SEE) will be used to improve energy efficiency, performance, security, and reliability. Data center capable abstraction of the underlying heterogeneity of the server is provided by an OpenStack-based middleware. In this chapter, we focus in particular on the architecture of the server platform including a dedicated high-speed, low latency communication infrastructure, give a short introduction into the software stack including thermal management strategies, and provide an overview of the targeted applications.

Proactive Workload Management for Bare Metal Deployment on Microservers.

This paper introduces a concept for an energy-aware workload management (WM) for heterogeneous microserver environments. Its main focus is on highly dynamic service-driven workloads often coupled to user requests requiring fast response times. The WM is developed in scope of the M2DC (Modular Microserver Data Center) project, in which a new server generation of composable microservers is designed. Targeting an easy industrial applicability, the underlying middleware is based on a turnkey OpenStack platform. As part of that middleware, the WM makes use of workload/utilization and power data as well as corresponding (prediction) models to deploy applications on the most suitable microservers and temporarily shut down unused capacities, either proactively or reactively (in case of deviations from forecasts). The WM has been implemented and simulated within a virtual environment. However, the integration, refinement and evaluation on the new M2DC hardware is still work in progress.