Collects and learns system behavior to create a
coherent multi-layer view
of the entire system. It offers topology@scale, including all layers from storage application, through virtual platform, to physical layer.
Continuously captures system metrics and performs cross-layer analytics to
and quickly identify suspect sub-systems. This includes equipment failure, service unavailability, system overload, and more.
Acts as a
for the interconnected storage system and the supporting communication infrastructure. It covers covering the entire path from storage system to consumers, providing tools to quickly identify and diagnose issues.
Topology Acquisition, Data Collection, Model Creation
CogNETive collects real-time network information for both topology and traffic (flow) to capture a complete network picture.
CogNETive captures information at multiple communication layers, from physical connectivity and routing, to virtual network layers, to application-level interaction. All information is time-stamped to provide a full history for exploration and analysis; for example, identify trends. All information is persisted the ElasticSearch time-based database, which supports a rich query language.
CogNETive uses a collection of host-level agents and remote probes to gather real-time and comprehensive network information, with zero footprint inside guest VMs and containers. Capture granularity is dynamically controllable per collector, from high-level birds-eye system view to packet-level drill down. This allows adjusting the amount of resources spent on capturing for different use-cases; for example, system-wide information can be continuously collected with minimal resource overhead.
CogNETive can inject traffic into the network. Injected traffic can be captured back to provide detailed debugging of network behavior; alternatively, it can be used to create artificial load for performance testing.
Custom probes can be added to CogNETive to capture richer data for specific environments, such as a middleware platform or a particular application.
Identification and Prediction of Issues, Anomalies & Trends
CogNETive allows you to explore network topology and network traffic together, presenting network flow information in its topological context.
Multiple layers can be explored together, for example, to drill down from application level logical connectivity to network paths, or separately, for example to focus on a problem point. The exploration view provides orientation information, real-time telemetry, insights, and visual alerts, by augmenting flow metrics, rich metadata, and even results of network data analytics, on top of the topology view. This augmentation brings together topological and traffic data and makes the data more accessible and intuitively understandable; thus, data becomes more usable and more valuable.
Full network navigation is supported, with zoom capabilities, custom filtering, and topology based context, allowing you to explore and investigate your network with the right amount of information for your use cases. Dynamic and automatic context helps you focus on the problem areas and remove distractions.
Harness the power of "time-travel" to troubleshoot problems that occur as a result of configuration modification or changed usage patterns – interactively explore and view historical data to compare current and past behavior and state and understand the impact of changes.
The topology view also provides an intuitive control interface, allowing actions and queries on selected graphic elements. For example, performing a search on information related to a selected node, injecting traffic, or invoking fined-grained capture.
Anomaly Investigation, Issue Location, Debugging, Probing
CogNETive provides integrated data processing tools and preconfigured analytic algorithms to convert the raw network data into information and provide insights, smart monitoring, and diagnosis.
Integrated Grafana and Apache Spark services allow you to query, visualize, alert on, and understand your network metrics; create, explore, a nd share dashboards; and easily apply machine learning and big-data techniques to the captured network data. These tools simplify and enable operations@scale, allowing automatic sifting through the entire network to detect abnormal behavior, draw attention to important problems (avoiding exhaustive alerts), diagnose and prioritize, and accelerate troubleshooting.
Topology-based analysis, such as network path exploration, is combined with time-based signal processing to provide operational intelligence and better network understanding. Analysis results are used to augment the network view, showing focal context and visual ques.
Research in the Computing as a Service department targets the transformation of IT through cloud computing, while leveraging skills in quality, security, storage, systems, and software-defined environments.
IBM Research - 2018 | CogNETive.email@example.com