In this episode, Toby and Krisztian take a hands-on dive into what it actually means to run your own data centre. Drawing on years of real-world experience racking servers, pulling fibre, configuring BGP and managing colocation suites, they walk through physical layout, cooling, power redundancy, network topology, capacity planning and hidden operational costs — then contrast all of this with the cloud. A rare and genuinely technical episode for anyone curious about what happens behind the abstraction layer.

Chapters

0:00 Introduction

1:27 Our data centre backgrounds

4:09 What does a data centre actually look like?

6:27 Tiers of availability

7:14 Racks, blades and colocation

13:06 Power redundancy: dual circuits and hot-swappable components

15:55 Network: SFPs, fibre optic and bandwidth design

18:59 BGP, ASN numbers and getting on the internet

24:13 Cooling: cold aisles and hot aisles

28:52 The cost breakdown: hardware, power, space, staffing

33:18 Capacity planning: the static nature of physical infrastructure

37:00 Team size and skills required

40:17 Hardware lifecycle and refresh cycles

44:05 How servers are ordered, received and racked

47:11 Should you run your own data centre?

50:10 Edge cases where on-prem makes sense

53:38 Hybrid cloud and AWS Outposts

57:09 Cloud vs data centre total cost of ownership

1:03:01 Environmental impact: waste heat and green data centres

1:07:24 How data centre skills transfer to the cloud

1:08:30 Summary: use the cloud if you can

Technologies Mentioned

AWS - https://aws.amazon.com

AWS Outposts - https://aws.amazon.com/outposts

Kubernetes - https://kubernetes.io

Juniper Networks - https://www.juniper.net