CloudFabrix announces Observability-in-a-Box with Edge AI Capabilities to simplify and accelerate AIOps deployments

November 10th, 2020

CloudFabrix is enhancing its AIOps platform with native Observability and AI at the edge capabilities to bridge the gap between Observability and AIOps solutions. Enterprises are struggling with unifying multitude of expensive monitoring deployments as well as gaps in observability, specifically for modern application architectures that include usage of microservices, containers and Kubernetes. CloudFabrix addresses these gaps by providing a simplified Observability-in-a-box solution that is built with best in class open source and open telemetry based components, all packaged up nicely as a single self-contained module that can be up and running within minutes. A unique aspect of the solution is the edge AI that is included in the box, to provide AI inference for all observability data closer to the source

These new capabilities enable IT organizations to future proof their monitoring and observability investments in a cost effective and vendor agnostic way by leveraging open source components, while also benefiting from reliability, customizations and enterprise class support offered by CloudFabrix. The Observability-in-a-Box solution runs in standalone mode or in conjunction with our AIOps platform, which is the preferred mode of deployment, as it provides all the AIOps benefits. Most of our customers want to embark on AIOps initiatives but have realized that they have visibility gaps to begin with. In such scenarios, customers start with implementation of our observability box and fill the visibility gaps. This lays a strong foundation for AIOps, after which the observability data is integrated with our AIOps solution to start realizing the promise of AIOps.

What is Included in the Observability-in-a-Box?

  1. Observability Unified Pack
    • Monitoring: Prometheus based performance monitoring visualized in Grafana
    • Events/Logs: Elasticsearch based event/log consolidation, visualized in Kibana
    • Traces: Open tracing for request tracing visualized in Jaeger
  2. EdgeAI: AI engine for model training and inferences
  3. EdgeCollector: Asset Discovery & Full Stack Mappings

I) Observability Unified Pack: Metrics, Logs & Traces:

This module enables customers to gain observability into IT workloads with metrics, events/logs and traces, aka 3-pillars of observability. A unique aspect of this module is that it combines multiple observability functions (Tracing, Monitoring, Log Collection) into one box, that are traditionally deployed as siloed solutions. For instance, with this module, customers can understand application behavior better, with end-to-end request tracing and pointing out the weakest link in the service chain, while also extrapolating system behavior with observation of logs and metrics during that time. This significantly enhances the value of observability by providing a complete picture for DevOps, SREs and ITops personnel.

Use Cases

Following are typical use cases and scenarios that can be addressed with this module:

  1. Application Performance Monitoring (APM): Application performance monitoring with request tracing 
    • Microservices based apps using Docker, Kubernetes etc.
    • Traditional n-tier apps
  2. IT Infrastructure Monitoring (ITIM): IT performance, availability and fault monitoring, alerting
    • Traditional and modern apps
    • Traditional IT infrastructure (compute, network, storage, firewall etc.)
    • Cloud native workload monitoring on AWS, Azure, Google Cloud
  3. Log Consolidation & Analytics 
    • Log Analytics – Syslogs, Web logs, Windows event logs, audit logs etc.
    • Network & Traffic Analytics – Netflow, PCAP etc.
    • Security Analytics (SIEM-like) – Firewall logs, IDP/IDS logs etc.

II) Edge AI

As more workloads continue to run on edge, IT organizations prefer to have intelligence built closer to the source of data, referred to as the edge. Traditionally ‘Edge’ term was used in relation to the IoT domain, but in the modern multi-cloud environment context, ‘Edge’ has broader meaning – which can be a customer edge (ex: managed customer), service provider edge, branch site, campus site or IOT edge.

Edge AI provides AI inferencing for workloads right at the source to enable augmented decision making. With this approach, AI models can be continuously trained and deployed without having to rely on a provider or backend for AI inferencing. This is especially useful in situations where data privacy, sovereignty and regulations require certain datasets to be kept within the boundary of enterprise

Edge AI provides following type of AI inferences:

  • Detecting anomalies in time series metrics
  • Detecting seasonality, baseline usage of metrics data
  • Detecting anomalies in trace duration
  • Predicting peak alert volume (alert volume converted as time series)
  • Predicting peak log volume (log volume converted as time series)

III) Edge Collector: Asset Discovery & Mapping

Typically, observability has focused on 3 key datasets: metrics, logs & traces, but through numerous customer implementations we have observed that asset dataset also plays a vital role in observability. Asset dataset provides visibility into assets, their configuration and the topology context. Observability tools focus more on proactive problem detection and behavior exploration leaving gaps in asset data, and an effective way to fill this gap is with a dedicated asset discovery & mapping capability that complements observability data. A key benefit of this approach is that IT personnel can now overlay observability data on an asset or vice-versa, providing a much richer operational context than it would have been without the asset discovery & mapping.

cfxEdgeCollector performs asset discovery and mapping using agentless mechanisms, like SNMP, SSH, WinRM/Powershell or using API integrations with element management systems (ex: vCenter, Cisco UCS, Microsoft SCOM etc.). Credentials are fully encrypted, stored locally and never leave the box, providing the additional layer of security for more regulated environments. Asset mapping is established by extrapolating dependency and topology from flow data, TCP connections, VM/host metadata, DNS lookups, MAC table etc.

Bringing everything together: Driving Outcomes and Advancing to AIOps with Observability-in-a-Box Data

Observability-in-a-Box fills the gaps in IT performance and analytics data and can serve as a quality data feed to AIOps platform, which is a key design principle of our overall AIOps solution. Observability-in-a-Box type solution has seamless integration with AIOps platform so that all metrics, events, logs, traces are available within the AIOps platform.

Benefits of integration Observability-in-a-Box data with AIOps

  • Seamless propagation of Observability data to AIOps pipeline
  • Aggregated full-stack insights and reporting
  • Define and track IT service Outcomes based on Observability data
  • Overlay observability data alongside asset data for richer insights.
  • Effective enrichment of alert/event data without many external dependencies – leading to effective correlation and noise reduction.

Getting Started is Easy

Getting started with Observability is easy and the solution can be up and running within minutes. Solution can be deployed on-premise, on cloud or on any existing Linux VM (CentOS/RHEL/OEL etc.) and the installer script will deploy all required software modules as containers. Data acquisition from apps can be setup using remote calls, APIs, exporters, beat modules and trace exporters. Monitoring and alerting behavior configuration is automated through Observability Profile.

Interested in learning more about this solution or for a free trial, visit

You might also like