Saturday, November 21, 2026

Top 5 This Week

Related Posts

Daniel Crowe, NETSCOUT: when excess data paralyzes AI – ZDNET


The quality of observability data today represents the nerve center of the success of AIOps (Artificial Intelligence for IT Operationseditor’s note). If companies have long considered that artificial intelligence (AI) should be powered by as much data as possible, this approach is now showing its limits.

However, when data proves to be fragmented, redundant, or devoid of operational context, its accumulation generates more noise than value.

In France, according to a recent ARCEP study, incoming data traffic is growing by more than 10% per year, fueled by the explosion of the cloud and micro-services. However, dumping this avalanche of raw information into AI systems does not create greater clarity, but rather tends to generate noise, confusion and an explosion in processing costs. Operational teams then find themselves drowning in unusable alerts, and the AI, unable to sort this incoherent flow, works blindly.

To get out of this paralysis, the solution involves better observability, rather than more substantial collection. By combining data through the dual lenses of network and security to eliminate data silos, blind spots and noise, companies eliminate many of the elements that risk compromising investments in AIOps.

When excess information harms visibility

Observability constitutes a central pillar in modern architectures. Microservices generate a significantly higher volume of telemetry than monolithic structures, and traditional rules-based alerts are no longer enough. Faced with this complexity, monitoring powered by AI is gradually becoming essential in companies. That being said, observability has different meanings depending on the role occupied.

Application teams approach it from the top down, tracking logs, metrics and traces across the different layers of the software stack. Network teams, on the other hand, approach it from the bottom up, observing actual traffic flow, service delivery, and user experience in real time. Despite the validity of both views, neither is complete or sufficient on its own.

The real problem lies in these perspectives coexisting in separate silos for years. Metrics on one side, logs on another, traces on a third, managing each fell to a different team and told an incomplete story. However, businesses can no longer analyze applications, services and networks independently of each other, nor examine them in isolation. Safety needs to be part of the conversation.

Ultimately, when companies seek to feed an AI system with siled and fragmented data, they are not providing it with a solid foundation for intelligent decision-making. They give him a puzzle with missing pieces.

Prioritize quality over quantity of data

In theory, when a problem arises, its resolution potential increases as more information is gathered about its causes. However, solutions other than accumulating more data exist. The costs of observability platforms can (and frequently do) exceed the cost of the target infrastructure itself. Add to this the consumption of tokens required by modern systems based on AI and large language models (LLM), and the economic equation quickly becomes unsustainable.

By practicing context optimization from the source, businesses can reduce redundant elements before they reach AI systems. This approach improves the relevance of the data that AI manages while helping to control token consumption and associated costs. Thus, leaders find true confidence in decision-making within their AIOps systems.

With data that is raw, lacking context, or sampled at the wrong time, AI cannot do its job. The result lacks precision, risks missing signals or sending teams chasing false positives rather than real problems. The result—more time spent, more money invested, and less confidence in the results—is the opposite of what AIOps promises.

Instead, businesses require “AI-ready” data with more context density. The objective is no longer to multiply records, but to produce information of real operational value. By supporting traditional logs, metrics, and traces with high-fidelity analysis from real-world network activity, businesses can provide their AI systems with a more complete and reliable understanding of the performance of their applications, services, and users. This way, they can break down silos and integrate network and application perspectives for complete visibility.

The pillars of “AI-ready” data

Truly AI-ready data is characterized by three inseparable pillars. It must first of all be directly observed, in particular from the reality of network packets, and not simply deduced, while being part of a continuous rather than sampled measurement. By leveraging concrete, real-time network activity, businesses can leverage Smart Data, structured, enriched, AI-ready metadata that depicts how applications, services and users perform across contexts.

This approach provides independent operational observability that complements traditional telemetry and gives AI systems a more reliable foundation for analysis and decision-making.

Then, it requires precise contextualization, i.e. a connection to the application, service or user experience that it describes. This context allows AI to weave meaningful connections rather than blindly searching for correlations amid raw noise. Finally, this data must simultaneously capture events occurring at the application layer, network and security environment.

Here, the convergence of network and security takes on its full meaning. A performance issue can be caused by application code, network infrastructure, storage, or a concurrent security incident. Deprived of a unified, high-fidelity vision encompassing all of these perimeters, the AI ​​works blindly, remaining capable of reporting an anomaly, but unable to identify the real cause.

Reconciling global visibility and budgetary control

The most successful AIOps teams recognize that the key to success goes beyond maximizing data volume. They require detailed observability that provides structured, credible and continuous intelligence, providing AI with the operational context needed for more reliable decisions. This development not only improves the operational resilience of companies, but also better controls the costs associated with observability and artificial intelligence.

As they explore AI-led operations, the quality of this context becomes even more consequential. AI systems that recommend or execute corrective actions require a reliable understanding of what is happening across the digital environment.

The question no longer becomes whether the AI ​​has enough data, but to guarantee their reliability, completeness, and contextualization to support a certain action.


The quality of observability data today represents the nerve center of the success of AIOps (Artificial Intelligence for IT Operationseditor’s note). If companies have long considered that artificial intelligence (AI) should be powered by as much data as possible, this approach is now showing its limits.

However, when data proves to be fragmented, redundant, or devoid of operational context, its accumulation generates more noise than value.

In France, according to a recent ARCEP study, incoming data traffic is growing by more than 10% per year, fueled by the explosion of the cloud and micro-services. However, dumping this avalanche of raw information into AI systems does not create greater clarity, but rather tends to generate noise, confusion and an explosion in processing costs. Operational teams then find themselves drowning in unusable alerts, and the AI, unable to sort this incoherent flow, works blindly.

To get out of this paralysis, the solution involves better observability, rather than more substantial collection. By combining data through the dual lenses of network and security to eliminate data silos, blind spots and noise, companies eliminate many of the elements that risk compromising investments in AIOps.

When excess information harms visibility

Observability constitutes a central pillar in modern architectures. Microservices generate a significantly higher volume of telemetry than monolithic structures, and traditional rules-based alerts are no longer enough. Faced with this complexity, monitoring powered by AI is gradually becoming essential in companies. That being said, observability has different meanings depending on the role occupied.

Application teams approach it from the top down, tracking logs, metrics and traces across the different layers of the software stack. Network teams, on the other hand, approach it from the bottom up, observing actual traffic flow, service delivery, and user experience in real time. Despite the validity of both views, neither is complete or sufficient on its own.

The real problem lies in these perspectives coexisting in separate silos for years. Metrics on one side, logs on another, traces on a third, managing each fell to a different team and told an incomplete story. However, businesses can no longer analyze applications, services and networks independently of each other, nor examine them in isolation. Safety needs to be part of the conversation.

Ultimately, when companies seek to feed an AI system with siled and fragmented data, they are not providing it with a solid foundation for intelligent decision-making. They give him a puzzle with missing pieces.

Prioritize quality over quantity of data

In theory, when a problem arises, its resolution potential increases as more information is gathered about its causes. However, solutions other than accumulating more data exist. The costs of observability platforms can (and frequently do) exceed the cost of the target infrastructure itself. Add to this the consumption of tokens required by modern systems based on AI and large language models (LLM), and the economic equation quickly becomes unsustainable.

By practicing context optimization from the source, businesses can reduce redundant elements before they reach AI systems. This approach improves the relevance of the data that AI manages while helping to control token consumption and associated costs. Thus, leaders find true confidence in decision-making within their AIOps systems.

With data that is raw, lacking context, or sampled at the wrong time, AI cannot do its job. The result lacks precision, risks missing signals or sending teams chasing false positives rather than real problems. The result—more time spent, more money invested, and less confidence in the results—is the opposite of what AIOps promises.

Instead, businesses require “AI-ready” data with more context density. The objective is no longer to multiply records, but to produce information of real operational value. By supporting traditional logs, metrics, and traces with high-fidelity analysis from real-world network activity, businesses can provide their AI systems with a more complete and reliable understanding of the performance of their applications, services, and users. This way, they can break down silos and integrate network and application perspectives for complete visibility.

The pillars of “AI-ready” data

Truly AI-ready data is characterized by three inseparable pillars. It must first of all be directly observed, in particular from the reality of network packets, and not simply deduced, while being part of a continuous rather than sampled measurement. By leveraging concrete, real-time network activity, businesses can leverage Smart Data, structured, enriched, AI-ready metadata that depicts how applications, services and users perform across contexts.

This approach provides independent operational observability that complements traditional telemetry and gives AI systems a more reliable foundation for analysis and decision-making.

Then, it requires precise contextualization, i.e. a connection to the application, service or user experience that it describes. This context allows AI to weave meaningful connections rather than blindly searching for correlations amid raw noise. Finally, this data must simultaneously capture events occurring at the application layer, network and security environment.

Here, the convergence of network and security takes on its full meaning. A performance issue can be caused by application code, network infrastructure, storage, or a concurrent security incident. Deprived of a unified, high-fidelity vision encompassing all of these perimeters, the AI ​​works blindly, remaining capable of reporting an anomaly, but unable to identify the real cause.

Reconciling global visibility and budgetary control

The most successful AIOps teams recognize that the key to success goes beyond maximizing data volume. They require detailed observability that provides structured, credible and continuous intelligence, providing AI with the operational context needed for more reliable decisions. This development not only improves the operational resilience of companies, but also better controls the costs associated with observability and artificial intelligence.

As they explore AI-led operations, the quality of this context becomes even more consequential. AI systems that recommend or execute corrective actions require a reliable understanding of what is happening across the digital environment.

The question no longer becomes whether the AI ​​has enough data, but to guarantee their reliability, completeness, and contextualization to support a certain action.

https://www.zdnet.fr/wp-content/uploads/zdnet/2026/11/daniel-crowe_netscout.jpg https://www.zdnet.fr/tribune/daniel-crowe-netscout-quand-lexces-de-donnees-paralyse-lia-508131.htm?utm_source=rss_feed&utm_medium=link&utm_campaign=unknown

Shredder Smith
Shredder Smith
Shredder Smith is the lead curator and digital persona behind topaitools4you.com, an AI directory dedicated to "shredding" through industry hype to identify high-utility software for everyday users. Smith positions himself as a blunt, no-nonsense reviewer who vets thousands of emerging applications to filter out overpriced "wrappers" in favor of tools that offer genuine ROI and practical productivity. The site serves as a watchdog for the AI gold rush, providing categorized rankings and transparent reviews designed to help small businesses and creators navigate the crowded tech landscape without wasting money on low-value tools.

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Popular Articles