Networking and Telecommunications for Data Centres

What you'll learn

• Explain the network layers, Ethernet frames, addressing and link speeds, and calculate link efficiency, frame rates, serialization delay and transfer times
• Break latency into its parts, size TCP windows from the bandwidth-delay product, estimate the effect of loss, incast, hashing and fan-out, and apply quality of service
• Design and size leaf-spine and five-stage Clos fabrics, calculating oversubscription, server counts, link and optic counts, and the effect of failures
• Explain how tenants are isolated with VRFs, overlays and host networking, and apply load balancing, segmentation, DDoS defence and encryption
• Compare block, file and object storage networks, size storage, rebuild and replication traffic, and relate RPO and RTO to replication design
• Describe switch and router hardware, calculate fabric size from radix and optics power, and specify airflow, power feeds and spares
• Describe how networks interconnect through transit, peering, exchange points and meet-me rooms, calculate 95th-percentile bills, latency and the availability of diverse routes
• Explain BGP path selection and route security, DNS and anycast, content delivery networks and enterprise WANs
• Select fibre types and optics, calculate optical loss budgets in dB and dBm, and explain DWDM, coherent optics and data centre interconnect
• Size −48 V DC plants, explain telecom facilities, access networks and NEBS, and calculate PTP offsets, holdover and satellite latency
• Explain the networks of an AI cluster, estimate collective communication times, size non-blocking fat trees and assess their facility impacts
• Operate networks safely with telemetry, automation and change management, recognize gray failures and cascading outages, and plan capacity