Bare metal in Kenya, Tanzania and Nigeria — without sending anyone

No hyperscaler runs a cloud region in Kenya, Tanzania or Nigeria. Amazon's African region is Cape Town; Microsoft's is in South Africa. So the default position for an international company serving these countries is a machine on another continent, and the default is what this product exists to change. A dedicated server in ke-1a, tz-1a or ng-1a puts the database, the ledger or the API in the same city as the people using it.

The distance you remove is larger than a map suggests. Paths between African networks have long been carried north to Europe and handed back rather than exchanged locally, and published measurement research puts intercontinental round trips from African networks at roughly 100 to 400 ms. A request from Lagos to Nairobi has commonly taken longer than the same request from Lagos to London. Cape Town is an international deployment from all three of these countries, not a regional one.

Each city carries its own connectivity underneath. Kenya's subsea capacity lands at Mombasa — TEAMS, SEACOM, EASSy, LION2, DARE1 and PEACE — and runs roughly 480 km inland to Nairobi, where KIXP has exchanged domestic traffic since 2002 across 141 peer networks and 2.9 Tbps of capacity; ke-1a peers there. Dar es Salaam is itself a landing point, with SEACOM, EASSy and SEAS ashore in the city and 13,820 km of state backbone fibre reaching the rest of Tanzania. Lagos takes eight subsea systems and interconnects Nigerian networks at IXPN, now 13 points of presence across seven states with more than 130 networks connected and peak domestic traffic past 2 Tbps by March 2026.

On a single-tenant machine the local advantage compounds, because nothing between you and the hardware is shared. Every core, every gigabyte of memory and every disk belongs to one workload. There is no hypervisor, no noisy neighbour on the same spindles, and no scheduling contention on the night your batch job matters. A machine in the city, uncontended, is about as short as the path from a Nairobi or Lagos user to your data gets.

This is the part of buying hardware abroad that nobody publishes, and it is the part that decides whether the deployment is a good idea. Start with delivery. The machine is ours, already in the building, already cabled and already on the network — there is no shipment, no clearing agent, no import duty assessed at a port, and no week spent waiting for a customs broker to release a pallet. Configurations held in stock come online within hours. A custom build is quoted with a short lead time while the exact hardware is assembled, burned in and tested, and that lead time is measured in days rather than in the six to ten weeks an international hardware purchase and shipment usually takes.

Then the console. Every machine ships with out-of-band IPMI or BMC access, which is the single most important thing on this page for a buyer with nobody in the country. It is a network path to the hardware that does not depend on the operating system being alive: power the machine off and on, watch it POST, mount an image, reinstall from scratch, change a BIOS setting, recover a box you locked yourself out of. Ninety-nine per cent of what you would have flown someone out to do, you do from your desk at two in the morning, in a browser.

Then the failures that IPMI cannot fix, which are the physical ones. A disk in a RAID-1 pair fails, a power supply goes, a stick of memory starts throwing correctable errors. On rented hardware that is our problem, not yours: replacement is included in the service and is carried out by engineers who are already in the building, and every machine ships with its storage in RAID-1 so a single disk failure is a degraded array rather than an outage. Compare that with the same failure on a machine you own in a colocation cage: a spare you have to have bought in advance and stored in-country, a courier, an access request, and someone on a plane if the spare turns out to be the wrong one.

Then everything else that needs hands. Recabling, a port move, a physical inspection, a stubborn power-cycle, a hardware swap you want witnessed — remote hands cover it, and you raise the request the way you raise any other ticket. What you should expect from us, and should demand in writing from anyone else quoting you, is the same short list: what remote hands includes, how you authorise someone to touch your machine, what the escalation path is out of hours, and what the uptime SLA pays out when it is missed. Ours is 99.9% with hardware replacement, and it is a term of the service rather than a line on a brochure.

Finally the clock. Nairobi and Dar es Salaam run on East Africa Time, UTC+3; Lagos on West Africa Time, UTC+1. Neither observes daylight saving, so the offset from your own office holds all year rather than shifting twice. From London that is one to three hours ahead; from New York, five to eight; from Singapore, five to seven behind. The practical consequence is that a European or Middle Eastern team overlaps the whole working day in all three countries, and a US East Coast team overlaps the morning.