19+ years in IT, telecom and corporate infrastructure — architecture, support and migrations at enterprise scale.
19+ years in IT, telecom and corporate infrastructure — 8 of them in pure telecom. Along the way: 50+ software implementation, support and migration projects, and 50+ projects architecting hardware around clients' solutions — owning enterprise architecture at, frankly, babysitter level, across dozens of vendors.
At peak: single-handed support for 45 server racks at once — 450 servers, 25 storage arrays, 20 switch pairs at a telecom operator. Support spanning both IBM Power and Oracle Exadata. Priority stack for the last five years: Lenovo + Nutanix, including solutions on Lenovo ThinkSystem and ThinkAgile. A certified Lenovo service engineer and a repeatedly-certified Nutanix specialist (NCP-MCI, NCP-US, NCP-DB, NCP-CN).
From the first Xeons and direct-attached LSI onward — hands-on support for gear from:
Designing and operating Nutanix.
VMware, OpenShift Virtualization and SUSE Virtualization.
Backup architecture on Commvault.
Hyperconverged storage; Lenovo / NetApp / Pure storage systems; Brocade FC SAN fabrics.
Infrastructure design for Kubernetes; database management on Nutanix.
Fortinet / Check Point, Juniper, Cisco, Arista — plus NVIDIA and Edgecore architecture.
End-to-end solution design, weak-point analysis of existing infrastructure, rebuilds, migrations and budget optimization.
Write-ups from practice: what really drives the budget and timeline of an infrastructure project, where the arithmetic parts ways with reality, and what to ask before a specification is signed. Every issue ends with a checklist.
Where an array fits, where local disks do, and where not to virtualise at all — the three forks that set the budget of a data platform.
Read the issue → Issue 02How the tape tier actually works, what breaks in it, and why LTO-10 changes the capacity arithmetic.
Read the issue → Issue 03Capacity, retention, backup window — and not a word about time to restore service. Where backup sizing parts ways with reality.
Read the issue → Issue 04A database complex between two halls and a two-rack office in a single week: why the plan never hinged on the rigging.
Read the issue → Issue 05How a licensing perimeter differs from storage capacity, and why the quoted reduction ratio dies on encrypted data.
Read the issue → Issue 06How much support to buy up front, why keep spares under an active contract, and when hardware stops deserving trust.
Read the issue → Issue 07Rent versus own across four environments at once: distance between sites, the price of memory, link arithmetic, and two boundaries for accelerators.
Read the issue → Issue 08The budget for an application system has four lines and the customer sees one. A sizing argument that nearly quadrupled the subscription, and the measurement that settled it.
Read the issue → Issue 09Five replacement options, and what happens to storage and networking once “ESXi plus an array over Fibre Channel” drops out of the design.
Read the issue → Issue 10Infrastructure is bought in cores and installed in kilowatts. The arithmetic of phases and feeds, and an estate where 71% of the bill is paid just for being on.
Read the issue → Issue 11What is actually in the racks: three node generations, a fifth of the power going to switches, and a temperature case where the obvious explanation proved wrong.
Read the issue → Issue 12Requirements grow faster than hardware depreciates. Five states of infrastructure debt, and the moment when there is nothing left to choose between.
Read the issue → Issue 13The data-protection ladder from a cloned VM to a copy in a local cloud: what each rung covers and which of it the platform licence includes.
Read the issue → Issue 14Where Kubernetes moves when you leave vSphere: five combinations of virtualisation and container platform, and the automation that survives the change.
Read the issue → Issue 15Cloud, colocation, rental and your own data centre: who covers which level of risk, what stays with you regardless, and why losing one building falls out of the registers.
Read the issue → Issue 16The world that is not going anywhere: VMware on subscription, a Fibre Channel fabric, tape, and a current-generation AI cluster in the same procurement queue.
Read the issue →Got an infrastructure project, a proof-of-concept, or a design review? Reach out on LinkedIn or by email.
