AI Weekly Malaysia

Vizient senior cloud engineer Dave Bradley recounted at VMware Explore how a bar encounter with senior execs led to a datacenter migration from Chicago to Dallas, initially proposed as a two-week deadline but settled at six-to-seven months. The team used Pure Storage (Everpure) arrays with 10Gbps inter-site links for replication, but faced mismatched IP schemes and insufficient hardware to run all workloads during downtime, forcing workload consolidation onto a subset of servers.

Why: The article is an entertaining war story with minimal actionable technical detail — it stops before explaining how the team actually solved the IP scheme mismatch or the consolidation approach. Skip it unless you want a cautionary tale about scope being set by execs in bars, not a migration playbook.

A 'Who, Me?' column shares a reader's confession about being sent as the 'Cisco expert' with only three months' experience to a customer running Nortel kit. After racking a new switch and configuring VLANs, his spanning-tree settings dropped over 4,000 AS/400 sessions, which the customer's IT team had to kill manually one by one because SNA-over-TCP/IP emulation couldn't tolerate a single lost packet.

Why: A relatable cautionary tale about imposter syndrome meets real-world blast radius: a single misconfigured spanning-tree setting on one switch took out thousands of production sessions. The practical lesson is to verify protocol-level tolerances (e.g., packet-loss sensitivity of legacy emulated sessions) before plugging new gear into a live network, and to push back when sent into unfamiliar territory.