(Source: Andriy Popov/Alamy)
A logjam is defined as "an immovable pileup or tangle of logs," such as those that accumulate in a river. On the network, which is its own communications river, network staff also encounter their share of logjams.
These pileups come in the form of drowning in an ocean of excessive logs. Excessive network logging clogs CPUs, overflows memory and confounds the network staff as they struggle to decipher which logs are – and should be – actionable.
Meanwhile, the daily logjams of data and workflows are morphing into a larger issue as network staffers strive to blend the tools that standard network monitoring, observability, AIOps, and now AI agents, have thrust upon them for monitoring telemetry and other network events at increasingly granular levels.
These technologies overlap with each other, and the duplication wastes corporate IT spend. How can IT gain cost controls? And how do network staff members avoid overlapping efforts when they are still struggling to understand which tools should be used for what?
Related:Is your network infrastructure ready for AI workloads?
IT networks today span central IT, edge locations, cloud locations, and remote home and field offices. The standard network monitoring tools that many sites still use were designed for monolithic networks, like a single corporate network in an enterprise. They can't handle the complexities of a hybrid network topology that goes beyond the walls of the enterprise.
Sites recognize this, and so do network vendors. Both see the need to update network management roadmaps, as almost no one is operating with a monolithic enterprise network anymore.
The questions are: What are the right tools and methodologies to upgrade to – and which existing tools can be eliminated?
The four categories of network monitoring and mitigation tools are:
Standard network monitoring stands on its own because it is a mature technology and sites have great familiarity with it. It uses metrics for network traffic, CPU and storage utilization, error allowances and response times, but IT must predefine these metrics. Monitoring tools issue alerts when these predefined metrics are exceeded, and then it is up to IT to find and resolve the issues.
Standard network monitoring falls short because it only reports on what IT predefines for it to report. Observability goes deeper. It will report when metrics are violated, but also where and why a violation occurred. It provides this information by looking at metrics, logs and traces –and the software can do this autonomously. This gives IT a head start in issue resolution.
Related:It's time to revamp IT security to deal with AI
The goal of AIOps is to extend the capabilities of observability with more AI and automation for issue resolution. The drawback of AIOps is that it has limited insight into the context of a network event when it analyzes data. It can't even tell if the telemetry data it is analyzing is good data. This is where IT must still step in because it takes network professionals to confirm the validity of what AIOps finds and to apply the fixes.
A new wave of AI network agent tools attempts to further automate the need for network staff to intervene in problem resolution. AI agents detect and resolve issues automatically. The agents do this by using machine learning to review network performance history so they can gain a business context of how the network should function.
Migrating from standard network monitoring to observability to AIOps and then to AI network agents is the natural progression of network management software. Companies and vendors see this, so an evolutionary roadmap of network management has been defined.
Related:AI, automation and IT layoffs: CIO guidance for managing disruption
But before they can get on the roadmap with these new technologies, companies must assess where they are along the path in terms of tools, staff, business requirements, and spend. Here are five best practices:
For many IT network staffs, sorting through current tools in use – and those that have been forgotten and on the shelf – is a prodigious task. But now is the time.
Network management tools should be inventoried on all networks throughout the enterprise, whether the networks are on prem in the data center, at enterprise edge locations, or in the cloud.
Tools should be classified by function, so that any tools that overlap are eliminated. If different tools are being used for the same functions at different network locations, these tools should be standardized to a single toolset. That simplifies the job for staff by clarifying which tools to use and train on.
Part of the tools inventorying and evaluation process is connecting with tool vendors to see where these vendors are headed with their tool roadmaps.
The roadmap for network management is clear: standard network monitoring to observability to AIOps to AI network agents.
If vendors don't have this evolution represented in their roadmaps, it's time to look for vendors that do.
Most corporate network staffs have a firm mastery of standard monitoring and are already engaged with observability.
The next step is introducing automation into observability through AIOps, which is still a work in progress because it requires a realignment and, in some cases, a reinvention of network work processes.
The network staff will be learning new AIOps tools, but also how to integrate additional AIOps automation into network workflows and daily operations.
These changes must be documented, and documentation is a weak area in network operations.
To ensure that operational documentation keeps pace with workflow changes, it's good practice to ask outside auditors to review documentation and operations so that any incongruences can be identified and corrected.
The concept of totally automated network operations using AI agents is still more theory than fact.
Nonetheless, some sites are dipping their toes into the water.
Network AI agents use machine learning to review past network performance so they can obtain a business context for their automation. But they don't have the hands-on know-how and expertise of a network employee.
A best practice is to initially deploy AI network agents on highly predictable and controlled networks that have a low risk of changes or anomalies.
Legacy technology not only means old – it also means tried, proven, and meant to last.
There will be network management tools that have stood the test of time and that continue to perform well.
When sites examine their tool benches, they should take a hard look at what continues to provide value.
By all means, upgrade tools and skills – but don't throw out what continues to work well.
President of Transworld Data
Mary E. Shacklett is an internationally recognized technology commentator and President of Transworld Data, a marketing and technology services firm. Prior to founding her own company, she was Vice President of Product Research and Software Development for Summit Information Systems, a computer software company; and Vice President of Strategic Planning and Technology at FSI International, a multinational manufacturer in the semiconductor industry.
Mary has business experience in Europe, Japan, and the Pacific Rim. She has a BS degree from the University of Wisconsin and an MA from the University of Southern California, where she taught for several years. She is listed in Who's Who Worldwide and in Who's Who in the Computer Industry.
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。