MacMusic  |  PcMusic  |  440 Software  |  440 Forums  |  440TV  |  Zicos
observability
Recherche

Overcoming modern observability challenges

lundi 23 décembre 2024, 10:00 , par InfoWorld
Observability, at its core, is about collecting and analyzing data generated from applications, cloud computing resources, and edge devices to ensure software and services are run smoothly, user experiences are optimized, and resources are managed cost-effectively. According to a recent Gartner report, 70% of companies that successfully apply observability will gain a competitive edge through faster problem identification and enhanced decision-making. 

However, as application development has progressed from monolith stacks to microservices, and user expectations have shifted toward instant digital experiences, companies have accumulated a mix of observability tools, each tailored to different parts of their tech stack. This hodgepodge raises costs, complicates data analysis, and slows issue resolution. 

The Kloudfuse observability platform addresses these challenges by building on three foundational principles: unification, cost optimization, and intelligent automation. Here’s a closer look at the observability landscape, the unique challenges organizations face today, and how Kloudfuse is designed to provide a unified, affordable, and intelligent approach to modern observability.

Observability challenge #1: Fragmentation and complexity

Traditionally, organizations have deployed multiple observability tools across their technology stacks to address distinct needs like monitoring logs, metrics, or traces. While these specialized tools excel individually, they rarely communicate well, resulting in data silos. This fragmentation prevents teams from gaining comprehensive insights, forcing devops and SRE (site reliability engineering) teams to rely on manual integrations to piece together a full picture of system health. The outcome is delayed insights and an extended mean time to resolution (MTTR), slowing down effective issue response.

Additionally, organizations now need to incorporate data streams beyond the traditional MELT (metrics, events, logs, and traces) framework, such as digital experience monitoring (DEM) and continuous profiling, to achieve comprehensive observability. DEM and its subset, real user monitoring (RUM), offer valuable insights into user interactions, while continuous profiling pinpoints low-performing code. Without integrating these data streams, teams struggle to link customers’ real experiences with specific code-level issues, resulting in data gaps, delayed issue detection, and dissatisfied customers.

Observability challenge #2: Escalating costs

The cost of observability has surged alongside the fragmentation of tools and the growing volume of data. SaaS-based observability solutions, which manage data ingestion, storage, and analysis for their customers, have become particularly expensive, with costs quickly accumulating. According to a recent IDC report, nearly 40% of large enterprises view high ownership costs as a major concern with observability tools, with the median annual spend by large organizations (10,000+ employees) on AIops and observability tools reaching $1.4 million.

Additionally, data transfer fees—commonly known as egress fees—have been a significant concern, as companies must pay for data leaving their platforms and entering SaaS observability clouds. When combined with pay-per-use pricing models, this often leads to unexpected budget overruns for enterprises. It’s no surprise that businesses are becoming increasingly frustrated with high SaaS observability costs, vendor lock-in, and the lack of deployment flexibility.

Observability challenge #3: Intelligence and insight

Observability is crucial for quickly detecting issues and taking corrective actions to ensure that application performance does not negatively impact customer experience. With millions of transactions occurring every second, relying on traditional logic, predefined rules, and human intervention is no longer sufficient. According to a 2023 Gartner report, applied observability has emerged as one of the top 10 strategic technology trends, underscoring the increasing need for using AI to make smarter, more automated solutions to stay competitive​ and optimize business operations in real time. 

Today’s observability solutions must go beyond static monitoring by incorporating AI and machine learning to detect patterns, trends, and anomalies. By automatically identifying outliers and emerging issues, AI-driven systems reduce the mean time to detect (MTTD) and mean time to resolve (MTTR), driving efficiency and helping teams address potential problems before they affect end-users.

A unified, cost-effective, and intelligent platform

Recognizing these challenges, Kloudfuse set out to build an observability platform that addresses the specific needs of modern cloud-native applications and dynamic digital experiences.

Unified observability

Kloudfuse brings together diverse data streams in a single platform, eliminating the need for multiple observability tools and manual data stitching. By unifying metrics, events, logs, and traces in its observability data lake, Kloudfuse empowers teams with a full-stack, unified approach to observability, enabling seamless correlation of front-end and back-end issues to significantly reduce MTTR.

With Kloudfuse 3.0, the observability data lake extends to integrate traditional MELT data with digital experience monitoring and continuous profiling, giving enterprises a comprehensive view of both user experience and system performance.

Real user monitoring delivers in-depth visibility into user interactions across every digital transaction—from front end to back end—enhanced with pixel-perfect replays of user journeys, providing full context for troubleshooting. Continuous profiling enables granular, line-level breakdowns of resource hot spots that cause application delays. This helps developers quickly debug and resolve performance bottlenecks for more efficient and reliable code. Additionally, continuous profiling identifies resource issues, such as CPU utilization and memory allocation, helping to reduce costs—a pillar of Kloudfuse discussed in the following section.

Cost optimization

Kloudfuse is designed with affordability and flexibility at its core. Unlike SaaS solutions that charge for data overages, Kloudfuse offers versatile deployment options, including private deployment in Amazon Web Services, Microsoft Azure, and Google Cloud Platform environments. This versatility not only avoids the high fees typical of SaaS and eliminates egress and data transfer fees, but also grants greater control over data residency—essential for organizations with strict data sovereignty requirements. Kloudfuse 3.0 expands the platform’s VPC private deployment model to support Arm processors, ensuring even further cost reduction and efficiency required by large-scale observability deployments. 

The Kloudfuse platform offers additional cost-saving features, such as patent-pending fingerprinting technology that automatically detects patterns in log messages to reduce storage and processing requirements, driving down associated costs. In Kloudfuse 3.0, the log archival and hydration feature allows customers to store logs more affordably using compressed JSON formats within the customer’s own storage, such as Amazon S3, meeting compliance standards without excessive expenses.

Additionally, Kloudfuse 3.0 includes cardinality analysis, metric roll-ups, and data shaping and transformation capabilities that convert high-volume data into valuable attributes, further reducing both storage and analysis costs. 

Intelligent automation

Kloudfuse’s integrated AI and machine learning capabilities enable teams to detect anomalies, trends, and root causes quickly. Advanced algorithms like rolling quantile, SARIMA, DBSCAN, seasonal decomposition, and Pearson correlation provide smarter alerting by pinpointing outliers, forecasting potential issues, and correlating observability data across streams to more quickly identify dependencies and root causes. The platform’s analytics tools also allow teams to visualize service dependencies and relationships with interactive service maps and topology views.

Kloudfuse 3.0 introduces Prophet for anomaly detection and forecasting to provide more accurate results, managing irregular time series that include missing values, such as gaps from outages or low activity, resulting in less tuning and improved forecasting, even with limited training data.

Kloudfuse 3.0 also introduces features like K-Lens and FuseQL to further enhance incident response and investigation. K-Lens uses outlier detection to quickly analyze thousands of attributes within high-dimensional data, identifying those that cause specific issues. It then uses heat maps and multi-attribute charts to pinpoint the sources of these issues, accelerating debugging and incident resolution. FuseQL offers a more expressive query language than LogQL, enabling advanced log analysis with multi-dimensional aggregations, filters, and anomaly detection, and supporting complex alerting conditions.

The future of observability 

Organizations need an observability solution that is comprehensive, cost-effective, and intelligent. The Kloudfuse observability platform is designed to monitor modern cloud-native workloads while optimizing costs, offering insights into model performance and mitigating risks. Kloudfuse also leverages AI and machine learning to enhance data analysis, improving alerting, issue resolution, and performance optimization. Recognizing the need for AIops and LLMops, Kloudfuse is set to enable proactive monitoring of generative AI applications, ensuring reliability and business continuity.

As observability becomes a strategic asset, Kloudfuse provides a unified, scalable solution, expanding to support emerging use cases and harnessing AI to continually improve platform intelligence.

Pankaj Thakkar is co-founder and CTO of Kloudfuse.



New Tech Forum provides a venue for technology leaders—including vendors and other outside contributors—to explore and discuss emerging enterprise technology in unprecedented depth and breadth. The selection is subjective, based on our pick of the technologies we believe to be important and of greatest interest to InfoWorld readers. InfoWorld does not accept marketing collateral for publication and reserves the right to edit all contributed content. Send all inquiries to doug_dineley@foundryco.com.
https://www.infoworld.com/article/3625613/overcoming-modern-observability-challenges.html

Voir aussi

News copyright owned by their original publishers | Copyright © 2004 - 2024 Zicos / 440Network
Date Actuelle
mar. 24 déc. - 02:03 CET