For Ukrainian businesses, power outages are a scenario you prepare for in advance, not during the first incident. When the power goes, what matters is not only how long the servers will last, but whether anyone can reach them, whether data stays intact and how quickly everything comes back.
Below is a practical plan for an IT manager or business owner: what to check, in what order, and how to avoid spending budget on protecting systems that can safely be stopped.
Start with one question: what must keep working without power
Protecting “everything” is expensive and usually unnecessary. First list your systems and answer two questions for each:
- How long can the system be down before it hurts the business? This is the Recovery Time Objective (RTO).
- How much data can you afford to lose? This is the Recovery Point Objective (RPO): the point in time to which data must be recovered after an outage.
Both terms are defined in NIST’s contingency planning guide for IT systems (SP 800-34 Rev. 1). In practice, systems fall into three groups:
| Group | Examples | What we do |
|---|---|---|
| Must work during an outage | accounting and payments, CRM, email, VPN, telephony | backup power, a second internet link or moving outside the office |
| Can wait a few hours | intranet portals, reporting, test environments | controlled shutdown and planned start-up |
| Can be stopped | archives, systems for one-off tasks | shut down first to extend battery runtime |
This table drives every later decision and the budget. Without it you cannot tell whether your current UPS, generator and internet links are enough.
Powering the server room: UPS, generator and shutdown order
A UPS buys minutes, not hours
An uninterruptible power supply (UPS) is there to bridge the switch to a generator or to shut systems down cleanly. Relying on batteries to carry the server room through a whole outage is rarely cost-effective.
What to check:
- UPS type. Servers and the network core need an online (double-conversion) UPS: the load is always fed through the inverter, so switching to battery and voltage spikes do not reach it. Line-interactive models are cheaper and fit less critical equipment.
- Real runtime. Calculate it from the actual load, not the datasheet. Battery capacity drops over time, so test runtime under load regularly.
- What is plugged in. Servers without power to switches, the router and the firewall keep running but cannot be reached. Network equipment belongs on the UPS just like the servers.
Generator and automatic transfer switch
Long outages need a generator with an automatic transfer switch (ATS). In this setup the UPS covers the gap between the grid going down and the generator starting, and smooths the generator’s power. Test the generator under real load, not only idling, and keep a fuel reserve and a delivery arrangement.
Power design is a job for electrical contractors. The IT team should give them accurate input: equipment consumption, the list of critical systems and the acceptable switchover time.
Controlled shutdown and start-up
If the batteries run low and there is no generator, or it fails to start, systems must shut themselves down in the right order. An abrupt power cut can damage databases, file systems and disk arrays.
- Connect the UPS to the network through its management card and configure the shutdown agent on your virtualisation hosts (VMware, Hyper-V).
- Set the order: applications and databases first, then virtual machines, hosts and storage. The network goes last.
- Start-up runs in reverse: network, storage, hosts, databases, applications. Write down who checks that everything is back up.
Don’t forget cooling
The server room air conditioner is usually not on the UPS. The servers keep running on battery while the room heats up. Set up temperature monitoring with alerts and an automatic shutdown when the temperature exceeds the limit allowed by the equipment manufacturer’s documentation.
Connectivity: two independent links
During outages the first thing to disappear is often the internet rather than a server: the provider’s equipment between you and its network needs power too.
- Two providers with different physical routes. Ask each whether their nodes have backup power and for how many hours.
- A third link as a last resort — mobile or satellite connectivity for the most critical services.
- Automatic failover. A next-generation firewall (NGFW) monitors the links and moves traffic to a working one. For example, with FortiGate SD-WAN the routes over a failed link are removed and traffic goes through the other links. More in network security and NGFW.
- IP telephony. If phones are powered by a PoE switch, that switch must be on the UPS too.
Move what is critical out of the office
Better still, critical systems should not depend on the office having power at all.
- Commercial data centre. Your own hardware is hosted in a data centre with backup power and several internet links. The office then only connects to the systems.
- Private cloud. Servers combined into a cluster provide high availability: if a node fails, another one takes over its role. See private clouds.
- Public cloud and hybrid. Some workloads — email, files, backups, web services — can move to the cloud while what needs to stay close remains on-site. How to choose a platform and plan the move: cloud solutions.
- Remote work. If systems are reachable over VPN or cloud services, people work wherever there is power, and the office stops being a single point of failure.
Backups that survive an outage
Backup is the last line of defence when something goes wrong: an emergency power cut, a disk damaged by a voltage spike, an error during start-up.
- Follow the 3-2-1 rule: three copies of the data, on two different types of media, one of them off-site.
- Check backup integrity automatically and test restores regularly: a backup you have never restored from guarantees nothing.
- After every emergency power cut, check the state of databases, file systems and disk arrays.
How we build backup and monitoring: IT infrastructure.
Action plan and drills
Technology will not help without people who know what to do.
- Describe the scenario: who decides on a shutdown, who checks the start-up, how employees and clients are informed.
- Keep contacts and instructions where they are available without servers and corporate email — including on paper.
- Monitor availability, UPS status and internet links, with alerts to the phones of the people responsible.
- Run drills regularly — for example, a planned power-off at the weekend. That way weak spots show up before a real outage.
Blackout readiness checklist
- A list of systems with RTO and RPO, split into three groups
- Servers, storage and network equipment are on a UPS
- Real battery runtime under the current load is known
- Generator with ATS tested under load, fuel reserve in place
- Controlled shutdown configured in the right order, start-up order documented
- Server room temperature monitoring and an emergency shutdown threshold
- Two internet links on different routes with automatic failover
- Critical systems reachable remotely or moved out of the office
- Backups follow the 3-2-1 rule, restore tested
- An action plan, contacts outside corporate systems and a drill schedule
How Bereg-IT can help
We will assess which systems are critical for your business and show the weak spots: server power, connectivity, backups, monitoring. Based on the findings, we design and configure backup links, controlled shutdown, backup and monitoring and, if needed, move workloads to a private or public cloud. You can start with a consultation or an audit.
The first consultation is free. Send a request and an IT engineer will get in touch.
Sources
- NIST SP 800-34 Rev. 1, Contingency Planning Guide for Federal Information Systems (2010) — contingency planning, RTO and RPO: https://csrc.nist.gov/pubs/sp/800/34/r1/upd1/final
- NIST CSRC Glossary — definitions of Recovery Time Objective and Recovery Point Objective (source: SP 800-34 Rev. 1).
- Fortinet, FortiGate / FortiOS 7.0 Administration Guide — Link monitoring example (SD-WAN): https://docs.fortinet.com/document/fortigate/7.0.1/administration-guide/478384