Michael Maurer

Low level Metrics to High level SLAs - LoM2HiS framework: Bridging the gap between monitored metrics and SLA parameters in cloud environments

Cloud computing represents a novel on-demand computing approach where resources are provided in compliance to a set of predefined non-functional properties specified and negotiated by means of Service Level Agreements (SLAs). In order to avoid costly SLA violations and to timely react to failures and environmental changes, advanced SLA enactment strategies are necessary, which include appropriate resource-monitoring concepts. Currently, Cloud providers tend to adopt existing monitoring tools, as for example those from Grid environments. However, those tools are usually restricted to locality and homogeneity of monitored objects, are not scalable, and do not support mapping of low-level resource metrics e.g., system up and down time to high-level application specific SLA parameters e.g., system availability. In this paper we present a novel framework for managing the mappings of the Low-level resource Metrics to High-level SLAs (LoM2HiS framework). The LoM2HiS framework is embedded into FoSII infrastructure, which facilitates autonomic SLA management and enforcement. Thus, the LoM2HiS framework detects future SLA violation threats and can notify the enactor component to act so as to avert the threats. We discuss the conceptual model of the LoM2HiS framework, followed by the implementation details. Finally, we present the first experimental results and a proof of concept of the LoM2HiS framework.

Adaptive resource configuration for Cloud infrastructure management

To guarantee the vision of Cloud Computing QoS goals between the Cloud provider and the customer have to be dynamically met. This so-called Service Level Agreement (SLA) enactment should involve little human-based interaction in order to guarantee the scalability and efficient resource utilization of the system. To achieve this we start from Autonomic Computing, examine the autonomic control loop and adapt it to govern Cloud Computing infrastructures. We first hierarchically structure all possible adaptation actions into so-called escalation levels. We then focus on one of these levels by analyzing monitored data from virtual machines and making decisions on their resource configuration with the help of knowledge management (KM). The monitored data stems both from synthetically generated workload categorized in different workload volatility classes and from a real-world scenario: scientific workflow applications in bioinformatics. As KM techniques, we investigate two methods, Case-Based Reasoning and a rule-based approach. We design and implement both of them and evaluate them with the help of a simulation engine. Simulation reveals the feasibility of the CBR approach and major improvements by the rule-based approach considering SLA violations, resource utilization, the number of necessary reconfigurations and time performance for both, synthetically generated and real-world data. Highlights? We apply knowledge management to guarantee SLAs and low resource wastage in Clouds. ? Escalation levels provide a hierarchical model to structure possible reconfiguration actions. ? Case-Based Reasoning and rule-based approach prove feasibility as KM techniques. ? In-depth evaluation of rule-based approach shows major improvements towards CBR. ? KM is applied to real-world data gathered from scientific bioinformatic workflows.

Model based engineering of Pichia pastoris central metabolism enhances recombinant protein production

The production of recombinant proteins is frequently enhanced at the levels of transcription, codon usage, protein folding and secretion. Overproduction of heterologous proteins, however, also directly affects the primary metabolism of the producing cells. By incorporation of the production of a heterologous protein into a genome scale metabolic model of the yeast Pichia pastoris, the effects of overproduction were simulated and gene targets for deletion or overexpression for enhanced productivity were predicted. Overexpression targets were localized in the pentose phosphate pathway and the TCA cycle, while knockout targets were found in several branch points of glycolysis. Five out of 9 tested targets led to an enhanced production of cytosolic human superoxide dismutase (hSOD). Expression of bacterial β-glucuronidase could be enhanced as well by most of the same genetic modifications. Beneficial mutations were mainly related to reduction of the NADP/H pool and the deletion of fermentative pathways. Overexpression of the hSOD gene itself had a strong impact on intracellular fluxes, most of which changed in the same direction as predicted by the model. In vivo fluxes changed in the same direction as predicted to improve hSOD production. Genome scale metabolic modeling is shown to predict overexpression and deletion mutants which enhance recombinant protein production with high accuracy.

Induction without methanol: novel regulated promoters enable high-level expression in Pichia pastoris

BackgroundInducible high-level expression is favoured for recombinant protein production in Pichia pastoris. Therefore, novel regulated promoters are desired, ideally repressing heterologous gene expression during initial growth and enabling it in the production phase. In a typical large scale fed-batch culture repression is desired during the batch phase where cells grow on a surplus of e.g. glycerol, while heterologous gene expression should be active in the feed phase under carbon (e.g. glucose) limitation.ResultsDNA microarray analysis of P. pastoris wild type cells growing in glycerol-based batch and glucose-based fed batch was used for the identification of genes with both, strong repression on glycerol and high-level expression in the feed phase. Six novel glucose-limit inducible promoters were successfully applied to express the intracellular reporter eGFP. The highest expression levels together with strong repression in pre-culture were achieved with the novel promoters PG1 and PG6.Human serum albumin (HSA) was used to characterize the promoters with an industrially relevant secreted protein. A PG1 clone with two gene copies reached about 230% of the biomass specific HSA titer in glucose-based fed batch fermentation compared to a PGAP clone with identical gene copy number, while PG6 only achieved 39%. Two clones each carrying eleven gene copies, expressing HSA under control of PG1 and PG6 respectively were generated by post-transformational vector amplification. They produced about 1.0 and 0.7 g L-1 HSA respectively in equal fed batch processes. The suitability in production processes was also verified with HyHEL antibody Fab fragment for PG1 and with porcine carboxypeptidase B for PG6. Moreover, the molecular function of the gene under the control of PG1 was determined to encode a high-affinity glucose transporter and named GTH1.ConclusionsA set of novel regulated promoters, enabling induction without methanol, was successfully identified by using DNA microarrays and shown to be suitable for high level expression of recombinant proteins in glucose-based protein production processes.

The response to unfolded protein is involved in osmotolerance of Pichia pastoris

BackgroundThe effect of osmolarity on cellular physiology has been subject of investigation in many different species. High osmolarity is of importance for biotechnological production processes, where high cell densities and product titers are aspired. Several studies indicated that increased osmolarity of the growth medium can have a beneficial effect on recombinant protein production in different host organisms. Thus, the effect of osmolarity on the cellular physiology of Pichia pastoris, a prominent host for recombinant protein production, was studied in carbon limited chemostat cultures at different osmolarities. Transcriptome and proteome analyses were applied to assess differences upon growth at different osmolarities in both, a wild type strain and an antibody fragment expressing strain. While our main intention was to analyze the effect of different osmolarities on P. pastoris in general, this was complemented by studying it in context with recombinant protein production.ResultsIn contrast to the model yeast Saccharomyces cerevisiae, the main osmolyte in P. pastoris was arabitol rather than glycerol, demonstrating differences in osmotic stress response as well as energy metabolism. 2D Fluorescence Difference Gel electrophoresis and microarray analysis were applied and demonstrated that processes such as protein folding, ribosome biogenesis and cell wall organization were affected by increased osmolarity. These data indicated that upon increased osmolarity less adaptations on both the transcript and protein level occurred in a P. pastoris strain, secreting the Fab fragment, compared with the wild type strain. No transcriptional activation of the high osmolarity glycerol (HOG) pathway was observed at steady state conditions. Furthermore, no change of the specific productivity of recombinant Fab was observed at increased osmolarity.ConclusionThese data point out that the physiological response to increased osmolarity is different to S. cerevisiae. Increased osmolarity resulted in an unfolded protein response (UPR) like response in P. pastoris and lead to pre-conditioning of the recombinant Fab producing strain of P. pastoris to growth at high osmolarity. The current data demonstrate a strong similarity of environmental stress response mechanisms and recombinant protein related stresses. Therefore, these results might be used in future strain and bioprocess engineering of this biotechnologically relevant yeast.

SLA-Aware Application Deployment and Resource Allocation in Clouds

Provisioning resources as a service in a scalable on-demand manner is a basic feature in Cloud computing technology. Service provisioning in Clouds is based on Service Level Agreements (SLAs) representing a contract signed between the customer and the service provider stating the terms of the agreement including non-functional requirements of the service specified as Quality of Service (QoS), obligations, and penalties in case of agreement violations. On the one hand SLA violation should be prevented to avoid costly penalties and on the other hand providers have to efficiently utilize resources to minimize cost for the service provisioning. Thus, scheduling strategies considering multiple SLA parameters and efficient allocation of resources are necessary. Recent work considers various strategies with single SLA parameters. However, those approaches are limited to simple workflows and single task applications. Scheduling and deploying service requests considering multiple SLA parameters such as amount of CPU required, network bandwidth, memory and storage are still open research challenges. In this paper, we present a novel scheduling heuristic considering multiple SLA parameters for deploying applications in Clouds. We discuss in details the heuristic design and implementation and finally present detailed evaluations as a proof of concept emphasizing the performance of our approach.

In Pichia pastoris, growth rate regulates protein synthesis and secretion, mating and stress response

Protein production in yeasts is related to the specific growth rate μ. To elucidate on this correlation, we studied the transcriptome of Pichia pastoris at different specific growth rates by cultivating a strain secreting human serum albumin at μ = 0.015 to 0.15 h–1 in glucose-limited chemostats. Genome-wide regulation revealed that translation-related as well as mitochondrial genes were upregulated with increasing μ, while autophagy and other proteolytic processes, carbon source-responsive genes and other targets of the TOR pathway as well as many transcriptional regulators were downregulated at higher μ. Mating and sporulation genes were most active at intermediate μ of 0.05 and 0.075 h–1. At very slow growth (μ = 0.015 h–1) gene regulation differs significantly, affecting many transporters and glucose sensing. Analysis of a subset of genes related to protein folding and secretion reveals that unfolded protein response targets such as translocation, endoplasmic reticulum genes, and cytosolic chaperones are upregulated with increasing growth rate while proteolytic degradation of secretory proteins is downregulated. We conclude that a high μ positively affects specific protein secretion rates by acting on multiple cellular processes.

Enacting SLAs in clouds using rules

The emergence of Cloud Computing raises the question of dynamically allocating resources of physical (PM) and virtual machines (VM) in an on-demand and autonomic way. Yet, using Cloud Computing infrastructures efficiently requires fulfilling three partially contradicting goals: first, achieving low violation rates of Service Level Agreements (SLA) that define non-functional goals between the Cloud provider and the customer; second, achieving high resource utilization; and third achieving the first two issues by as few time- and energy consuming reallocation actions as possible. To achieve these goals we propose a novel approach with escalation levels to divide all possible actions into five levels. These levels range from changing the configuration of VMs over migrating them to other PMs to outsourcing applications to other Cloud providers. In this paper we focus on changing the resource configuration of VMs in terms of storage, memory, CPU power and bandwidth, and propose a knowledge management approach using rules with threat thresholds to tackle this problem. Simulation reveals major improvements as compared to recent related work considering SLA violations, resource utilization and action efficiency, as well as time performance.

LAYSI: A Layered Approach for SLA-Violation Propagation in Self-Manageable Cloud Infrastructures

Cloud computing represents a promising computing paradigm where computing resources have to be allocated to software for their execution. Self-manageable Cloud infrastructures are required to achieve that level of flexibility on one hand, and to comply to users requirements specified by means of Service Level Agreements (SLAs) on the other. Such infrastructures should automatically respond to changing component, workload, and environmental conditions minimizing user interactions with the system and preventing violations of agreed SLAs. However, identification of sources responsible for the possible SLA violation and the decision about the reactive actions necessary to prevent SLA violation is far from trivial. First, in this paper we present a novel approach for mapping low-level resource metrics to SLA parameters necessary for the identification of failure sources. Second, we devise a layered Cloud architecture for the bottom-up propagation of failures to the layer, which can react to sensed SLA violation threats. Moreover, we present a communication model for the propagation of SLA violation threats to the appropriate layer of the Cloud infrastructure, which includes negotiators, brokers, and automatic service deployer.

Cost-benefit analysis of an SLA mapping approach for defining standardized Cloud computing goods

Due to the large variety in computing resources and, consequently, the large number of different types of service level agreements (SLAs), computing resource markets face the problem of a low market liquidity. Restricting the number of different resource types to a small set of standardized computing resources seems to be the appropriate solution to counteract this problem. Standardized computing resources are defined through an SLA template. An SLA template defines the structure of an SLA, the service attributes, the names of the service attributes, and the service attribute values. However, since existing research results have only introduced static SLA templates so far, the SLA templates cannot reflect changes in user needs and market structures. To address this shortcoming, we present a novel approach of adaptive SLA matching. This approach adapts SLA templates based on SLA mappings of users. It allows Cloud users to define mappings between a public SLA template, which is available in the Cloud market, and their private SLA templates, which are used for various in-house business processes of the Cloud user. Besides showing how public SLA templates are adapted to the demand of Cloud users, we also analyze the costs and benefits of this approach. Costs are incurred every time a user has to define a new SLA mapping to a public SLA template due to its adaptation. In particular, we investigate how the costs differ with respect to the public SLA template adaptation method. The simulation results show that the use of heuristics within adaptation methods allows balancing the costs and benefits of the SLA mapping approach.