CBL - Campus del Baix Llobregat

Projecte llegit

Títol: Modernización del sistema de supervisión mediante telemetría activa y el stack TIG

Estudiants que han llegit aquest projecte:

Director/a: CASTAN PONZ, JOSÉ ANTONIO

Departament: DEGD

Títol: Modernización del sistema de supervisión mediante telemetría activa y el stack TIG

Data inici oferta: 06-10-2025     Data finalització oferta: 06-05-2026


Estudis d'assignació del projecte:
    DG ENG AERO/TELEMÀT
Tipus: Individual
 
Lloc de realització: Fora UPC    
 
        Supervisor/a extern: Maria Isabel Gandia Carriedo
        Institució/Empresa: CSUC
        Titulació del Director/a: Enginyera Superior de Telecomunicació
 
Paraules clau:
Telemetría activa, Stack TIG (Telegraf, InfluxDB, Grafana), Anella Científica (CSUC), TWAMP (Two-Way Active Measurement Protocol / RFC 5357), IP SLA (UDP Jitter), Aislamiento multicliente (Multi-tenancy), Monitoreo proactivo
 
Descripció del contingut i pla d'activitats:
Aquest treball de recerca té com a objectiu analitzar i millorar el rendiment de xarxes de comunicacions en entorns crítics, on la latència i el jitter poden afectar el funcionament de serveis essencials.
Per fer-ho, s'utilitzaran dues eines avançades de monitoratge desenvolupades dins de la comunitat GÉANT: Timemap, que permet mesurar i visualitzar la latència entre punts distribuïts de la xarxa, i PerfSONAR juntament amb les sondes Atlas, que faciliten proves de rendiment i una visió externa de l'estat de la xarxa.
El projecte planteja una possible implantació d'aquestes eines a l'Anella Científica, amb l'objectiu de detectar colls d'ampolla, millorar la fiabilitat de les comunicacions i optimitzar la qualitat dels serveis en temps real.
 
Overview (resum en anglès):
This project addresses the comprehensive modernization of the supervision system for the CSUC's Anella Científica (Scientific Ring), a critical infrastructure that connects the academic and research community in Catalonia. The primary objective is to evolve from a traditional statistical monitoring model-limited by a reactive approach and low sampling frequencies-toward an advanced telemetry system capable of proactively auditing the actual traffic experience. The solution aims to provide scalable multi-tenant visibility for over one hundred institutions, obtaining accurate metrics for parameters such as latency and jitter, both of which are fundamental for high-performance research.
The methodology employed is based on the design and deployment of the TIG technology stack (Telegraf, InfluxDB, and Grafana), orchestrated using Docker containers to guarantee the portability and scalability of the service. To validate the technical feasibility before moving to production, a virtual laboratory was built in the EVE-NG environment, faithfully replicating the hierarchical and redundant topology of the real network. Metric collection is carried out via a hybrid approach: by implementing custom Python probes, the system queries the core routers through SNMP to extract IP SLA operation data and, simultaneously, triggers active measurements using the standard TWAMP protocol (RFC 5357). Furthermore, a Python-based automation engine was developed to handle the bulk provisioning of clients from a master file, enabling the programmatic and efficient creation of isolated buckets, security tokens, and visual environments.
The results obtained confirm that the limitations of current monitoring tools are successfully overcome. A proactive architecture has been validated, enabling the detection of asymmetries and one-way link degradations by leveraging hardware-based timestamping. The system guarantees the sovereignty and privacy of institutional data through a strict multi-tenant isolation model based on restricted access tokens. In terms of operational efficiency, the solution has proven to be highly scalable, with optimized storage consumption that would allow managing one hundred institutions with only twelve gigabytes of data per year. In conclusion, the project delivers a functional and highly automated prototype that significantly reduces costs compared to proprietary commercial solutions, offering an accurate monitoring platform ready for deployment in real production environments.

© CBLTIC Campus del Baix Llobregat - UPC