Situation-Aware IT Asset Management

IT Asset Documentation counts to the centrepieces of most IT management tasks and processes such as risk management, IT architecture planning, change and incident management. In corporate practice, IT asset documentation is typically materialized within Enterprise Architecture Management (EAM) tools or Configuration Management Data Bases (CMDBs), capturing assets on infrastructure, application and business level and their interdependencies. Today, there is strong indication that current EAM tools and CMDBs are not capable to meet the requirements of a new generation of dynamically evolving, highly interrelated IT assets emerging in scenarios such as smart grids, smart factories or cloud service hosting. Several recent surveys revealed that users rate the data maintained in EAM tools as outdated, at an inappropriate level of granularity or incomplete and inconsistent. The TxtureSA project proposal is part of our longstanding endeavour to explore the causes of these quality issues and to develop novel cost-effective techniques and tools to raise the quality of IT asset documentation. Within TxtureSA we will focus on a widely neglected aspect in research and practice so far: the enhancement of IT asset models by information drawn from external data sources such as the runtime environment. Our goal is to take a significant step towards situation-aware knowledge bases supporting IT managers and IT architects to take informed decisions. In order to achieve this, two major challenges have to be solved. First, Functional Challenges concern the selection of appropriate data sources, visualizations and data propagation rules to spread metrics along the asset dependencies. Second, Repository Challenges concern the lifecycle support of metrics enhanced IT asset models ranging from metrics collection configuration languages, performant storage and queries, and appropriate versioning mechanisms. Within TxtureSA we will both conduct scientific profound empirical studies with industrial experts to solve the functional challenges and develop a prototypic generic model repository solving the repository challenges. The results will be validated in two case studies within the context of Green IT and Security Risk Assessment.

Information Technology has become a critical success factor in modern business across all domains. Software applications act as the backbone for companies, ranging from self-hosted services for in-house use to globally employed and highly available public services such as E-Commerce platforms and messaging systems. Major companies employ their own data centers to operate these systems, while others are moving their services into the public cloud, or opt for a hybrid approach. In every scenario, the ability to keep track of the dependencies between business services and applications is crucial e.g. for traceability, auditing and planning purposes. Furthermore, with the ubiquitous usage of virtualization technologies in data centers, applications no longer run directly on hardware, but are instead deployed on virtual machines. These virtual machines may belong to clusters, load balancers and other orchestration mechanisms. The resulting deployment stack of an application can therefore become very complex and span multiple layers of abstraction. Traceability is a key concern in this area as well. The ability to quickly identify the root cause of an application that went offline can make the difference of several days of downtime compared to a couple of minutes. Conversely, determining which applications depend on which physical machine can aid operations managers in detecting bottlenecks and optimizing resource utilization. The structural IT landscapes today are documented in several different sources, ranging from spreadsheet documents to Configuration Management Databases (CMDBs), which greatly complicates global analysis. Aside from the structural aspects, there is also the need to take runtime data into account. For instance, monitoring systems allow to determine which individual machines are online. However, they only offer rudimentary support for more sophisticated aggregated metrics. In the course of the txtureSA project we developed concepts and a tool prototype to collect data from several heterogeneous sources and represent this information in a single, versioned model. This structural model is enriched with runtime information to allow for highly refined analysis. With our flexible change propagation, we allow model changes to trigger further changes. This mechanism can be used to determine the runtime status of a business service by inferring it from the runtime status of the physical machines on which it transitively depends. In addition, our repository transparently maintains a full version history of all changes, allowing for queries over time.