APC by Schneider Electric · Egypt
The part nobody thinks about until the power goes.
APC UPS, rack PDUs and power distribution, sized properly, installed inside the racks we build, and maintained on a schedule rather than replaced after a failure.
What we deliver
UPS
From single-rack units to data-centre grade systems, sized against real measured load and the runtime you actually need to shut down or fail over cleanly.
Rack power
Metered and switched PDUs, dual power feeds to dual-PSU equipment, and distribution designed so a single failure does not take a rack with it.
Monitoring & maintenance
Network cards, alerting into the monitoring we already run, and scheduled battery inspection and replacement.
Sizing is where this goes wrong
Measured, not estimatedNameplate ratings on equipment are what it could draw, not what it does. We size on measured load with headroom for growth.
Runtime stated honestly“Thirty minutes” is a function of load. At full load the same unit may give you eight. We will tell you the number for your load, not the marketing one.
Batteries ageUPS batteries degrade from the day they are installed and typically need replacing every three to five years. An untested UPS is often the single point of failure in an otherwise redundant room.
Dual feeds used properlyDual-PSU servers plugged into the same PDU are not redundant. It is a common and expensive mistake, and we check for it on every takeover.
What we actually calculate, specify and check
Power is the only layer where every other redundancy you paid for can fail at once. Clustered nodes, dual controllers and mirrored storage all stop together on a single power path, and the failure usually traces back to a decision taken in ten minutes at purchase time. These are the calculations we do instead.
Sizing, measured, not estimated
Real load, meteredNameplate ratings tell you what equipment could draw at maximum, not what it draws in service. We meter the actual load over a period that includes the busy hour before specifying anything.
VA and watts are not the same numberA UPS is rated in both, and equipment specified against the wrong one is under-sized. We size against both figures and against the power factor of the actual load, which for modern servers behaves differently from older equipment.
Headroom for growthA UPS sized exactly to today’s load is obsolete the day a server is added. We size with deliberate headroom, and note that most units are also least efficient when running very lightly loaded, so the headroom is chosen, not maximised.
Inrush at start-upEquipment draws considerably more at the moment it powers on. A UPS that carries the running load comfortably can still trip when everything restarts together after an outage. Restart order matters and gets planned.
Redundancy stated plainlyWhether the room is designed to survive losing a UPS module, and what “N+1” actually means for your load. It is a specification decision with a price, not a word in a brochure.
Which topology, and why it matters
| Topology | How it behaves | Where it belongs |
|---|---|---|
| Line-interactive | Runs the load from the mains, regulates moderate voltage variation, and switches to battery on failure with a brief transfer. | Individual workstations, small comms cabinets, edge equipment with tolerant power supplies. |
| Double-conversion online | The load always runs from the inverter, so mains disturbances never reach it and there is no transfer time at all. | Server rooms, data centres, medical and payment equipment, and anywhere the incoming supply is unstable, which in much of Egypt it is. |
| Modular / scalable | Power and battery modules added as load grows, with redundancy inside the frame and modules replaceable without dropping the load. | Rooms expected to grow, and rooms where a maintenance window is genuinely hard to obtain. |
Runtime and batteries, the part that quietly expires
Runtime is a curve, not a numberA unit quoted at thirty minutes gives that figure at a specific load. At full load the same unit may give you single-digit minutes. We state the number for your measured load.
What the runtime is actually forDecide first: ride through short outages, or shut down cleanly, or bridge to a generator. Each needs a different runtime, and buying an hour of battery when you needed four minutes and a generator is money spent in the wrong place.
Batteries degrade from day oneSealed lead-acid batteries typically need replacing every three to five years regardless of whether they have ever been used. A UPS with original batteries after six years is a single point of failure wearing a green light.
Heat halves battery lifeBattery life falls sharply above room temperature. A UPS in a hot cupboard will not reach its rated life, and this is the most common reason batteries fail early in Egyptian installations.
Self-test and load testingAutomatic self-tests scheduled, and a real load test at a planned time. A self-test that has never been reviewed is not evidence of anything.
Replacement budgeted, not discoveredInstall dates recorded and replacement scheduled ahead of failure, so it is a planned purchase rather than an emergency one during an outage.
Distribution inside the rack
Dual feeds used properlyDual-power-supply equipment plugged into two independent PDUs on two separate paths. Both cords into the same strip is not redundancy, and it is the single most common fault we find on a takeover, usually on the most important server in the room.
The right class of PDUBasic distribution where nothing more is needed; metered where you need to know the real load per rack; outlet-switched where remote power-cycling saves a site visit at three in the morning.
Per-rack load balancedLoad distributed across phases and feeds rather than concentrated, and each rack’s draw known rather than assumed.
Breaker coordinationProtection sized so a fault in one rack trips its own breaker and not the one upstream feeding the room.
Outlet types and counts checkedThe correct connector types in the correct quantity for the equipment actually going in, planned before installation day rather than solved with adaptors afterwards.
Cable management and labellingEvery power cord traced and labelled at both ends. Half of all accidental outages are somebody unplugging the wrong thing with complete confidence.
Monitoring, shutdown and the room around it
Network management cards, alerting to usBattery health, load, input quality and events reported into the monitoring we already run, so a degraded battery or a failed input phase is a ticket, not a discovery.
Graceful shutdown that has been testedShutdown agents on the hypervisors, storage and network so equipment powers down in the right order before the battery is exhausted. Then actually tested, because an untested shutdown sequence usually has one machine in it that never received the signal.
Generator interactionThe UPS has to bridge the gap until the generator starts and stabilises, and it has to tolerate the generator’s output once it does. Both are checked, not assumed.
The UPS is also a heat sourceIts losses become heat in the same room it is protecting. Cooling load is calculated including the power equipment, not just the servers.
Emergency power off, and its risksWhere an EPO circuit exists, it is documented, guarded and tested deliberately, an unlabelled button next to a light switch is a real and recurring cause of unplanned outages.
What maintenance actually looks like
| Activity | How often | Why it matters |
|---|---|---|
| Remote health and alarm review | Continuous, as part of the managed service | Catches a failed battery string or a lost input phase the day it happens rather than at the next outage. |
| Physical inspection: terminals, fans, filters, temperature | Every six to twelve months | Loose terminals and blocked airflow are found by looking, not by a management card. |
| Battery condition assessment | Annually, more often past year three | Capacity falls long before the unit reports a fault. |
| Load test and shutdown rehearsal | Annually, at a planned time | The only way to know the runtime and the shutdown sequence are real. |
| Battery replacement | Typically every three to five years | Planned and budgeted rather than emergency-purchased mid-incident. |
Power is the layer that makes every other redundancy real. Dual controllers, clustered nodes and mirrored storage all fail together on a single power path.
Ask us to check what your current UPS would actually give you
It is a short exercise and the answer is often uncomfortable. Part of the free assessment, no obligation.
Book the free assessment02 35375791