How Andisa IT’s MikroTik consultants turned a second leased line into automatic failover for a rural Scottish wireless ISP with around 350 PPPoE customers and critical wind farm monitoring traffic.
Written by Andy Morrison, Managing Director, Andisa IT. Technically reviewed by Andy Morrison, Managing Director. Published 21/10/26. Last updated 21/10/26.
IN BRIEF
DS Electrical Angus had two leased lines but no automatic failover, so a circuit failure meant manually reconfiguring hundreds of PPPoE connections. Andisa IT rebuilt the core MikroTik network in a test lab, proved the design, then deployed it live. Leased-line failover is now automatic and typically causes around five seconds of disruption. The network came through the winter storms of 2024/25 with no unplanned customer service loss beyond physical mast failures.
PROJECT AT A GLANCE
| DETAIL | SUMMARY |
| Client | DS Electrical Angus Ltd, trading as Angus WISP |
| Type of network | Rural wireless internet service provider (wireless ISP) |
| Coverage | Around 600 miles of network, from Loch Tay towards Perth and the coast |
| Customers | Around 350 PPPoE customers, plus wind farm monitoring and control traffic |
| Core equipment | Two core MikroTik routers running RouterOS, wireless backbone with fibre where needed |
| Working with Andisa since | 2022 |
| Project | Automatic dual leased-line failover, network segmentation, secure remote management, automated backups |
| Configuration | Around 6,000 lines of MikroTik RouterOS across two core routers |
| Failover time | Typically around five seconds |
| Proven by | The storms of winter 2024/25 |
| Next stage | Segmented backbone with local PPPoE and Starlink failover for isolated sections |
The client: a 600-mile wireless ISP carrying critical traffic
DS Electrical Angus Ltd, trading as Angus WISP, runs a rural wireless internet service across a wide area of Scotland, from Loch Tay towards Perth and the coast. Founded and run by control engineer David Short, it connects around 350 PPPoE customers across roughly 600 miles of network infrastructure.
The backbone is mainly wireless, with fibre where businesses and organisations cannot be reached by line-of-sight links. MikroTik equipment sits at the core of the network and at key customer sites, alongside a mix of residential customer routers.
The network also carries communications used to monitor and control wind farm infrastructure. That makes reliability far more than a customer service issue.
The challenge: two leased lines, but no automatic failover
David designed and built the network himself. He had worked with several MikroTik specialists before approaching Andisa in 2022, and found us specifically for our MikroTik expertise.
DS Electrical had added a second leased line at the centre of the network for extra capacity and resilience. The capacity arrived. The resilience did not, because the existing architecture could not use both circuits for failover.
- Hundreds of PPPoE connections, many needing public addressing, made moving traffic between the two lines extremely complex.
- Failover was manual. If the primary line failed, David had to log in, reconfigure hundreds of PPPoE connections, force them to reconnect, then nurse systems with stale connections back into service.
- Failures hit at the worst time. Severe Scottish weather was the most likely cause of an outage, and the same weather often meant engineers were needed to keep generators running for remote villages.
It worked, but it was not scalable. Restoring internet connectivity could not keep turning into a major manual recovery operation.
How we redesigned a live ISP network without risking it
Changing the architecture of a live network that supports hundreds of users and operational systems carries real risk. So we did not experiment on the live system.
- Built a replica of the core network at Andisa’s Harrogate office.
- Designed the new architecture across the two core MikroTik routers, using around 6,000 lines of RouterOS configuration to manage hundreds of PPPoE connections across two separate leased lines.
- Tested and proved automatic failover in the lab, including how services recover when a circuit is lost.
- Deployed the proven configuration to the live network during a carefully selected quiet period.
The complexity was not there for its own sake. It was what it took to make intelligent use of two leased lines and let the network recover on its own when connectivity is lost.
How the automatic leased line failover works
Before the redesign, a leased-line failure meant manually reprogramming hundreds of PPPoE connections. Today, failover is automatic:
- A leased line fails. The core MikroTik routers lose the circuit.
- PPPoE sessions reconnect through the alternative leased line.
- Dynamic DNS updates automatically. Because the public IP address changes, each reconnection forces a Dynamic DNS update.
- DNS names stay the same. Customers and services keep using the same external DNS names, even though the connection underneath has changed.
- Services recover on their own. Monitoring systems reconnect, VoIP services re-authenticate and normal operation resumes.
A failure that once demanded significant manual work now typically causes around five seconds of disruption.
Keeping monitoring and control traffic running
Because the network carries wind farm monitoring and control communications, the design had to consider more than whether a connection is up. It had to consider how each system behaves when a component fails, and how it recovers without someone stepping in. That thinking comes from control engineering, a background David and Andisa share (see below).
Network segmentation: separating customer and management traffic
The project was also a chance to improve the structure of the network. We introduced:
- Separate customer and management VLANs, so network management traffic is kept apart from customer traffic.
- Segregation between individual customers, so one customer’s traffic cannot reach another’s.
This network segmentation improved security and made the network much easier to route and manage.
Secure remote management and automated backups
Before the redesign, reaching some management interfaces meant David travelling to a data centre and connecting locally before troubleshooting could even start. The new management network gives secure remote access to the infrastructure, so problems can be investigated without the journey.
We also introduced scheduled daily backups of network devices, giving a reliable recovery point if hardware or configuration is ever lost.
Tested by a Scottish winter
A resilience design only proves itself when something fails. During the winter of 2024/25, the network faced a succession of storms and severe weather, and the redesigned architecture kept operating.
Since the changes went in, DS Electrical has had no customer loss of service other than planned outages or complete physical mast failures caused by adverse weather.
“Andisa have improved my MikroTik setup to a level that provides me with 100% uptime, compliance and reliability, allowing me to concentrate on growing my business rather than fixing it. If issues do arise, I know that I can always rely on a quick turnaround and a competent resolution to the problem. Absolutely outstanding.”
David Short, Founder, DS Electrical Angus Ltd / Angus WISP
The next stage: Starlink failover for isolated network segments
Dual leased lines now protect the core. But in rural Scotland, a severe enough storm can physically cut off a section of the wireless backbone. Andisa and DS Electrical are developing the next stage of the resilience strategy:
- Smaller, more autonomous segments. The backbone is being divided into sections with their own local PPPoE services.
- Starlink failover for selected sections. If a central mast or backbone link fails, an isolated section can keep authenticating customers locally and route traffic through its own Starlink backup connection.
- Designed around real limits. Dedicated public IP addresses cannot be preserved in this emergency mode. Critical monitoring services mainly make outbound connections, so they keep sending data, and Dynamic DNS lets services such as VoIP re-register and stay operational.
The aim is not to pretend a 600-mile network exposed to Scottish weather can never suffer a physical failure. It is to make sure that when something does fail, as much of the network as possible carries on working.
The results
- Automatic leased-line failover, typically around five seconds of disruption, replacing manual reconfiguration of hundreds of connections.
- Two leased lines used for resilience, not just extra bandwidth.
- No unplanned customer service loss since implementation, other than complete mast failures caused by adverse weather.
- Around 350 PPPoE customers supported across roughly 600 miles of network.
- Customer, management and individual customer traffic segregated for security and manageability.
- Secure remote management removes unnecessary journeys to infrastructure sites.
- Daily automated configuration backups improve recoverability.
- A platform that can keep growing without David manually recovering hundreds of connections when something goes wrong.
Most importantly, David now spends less time fixing infrastructure and more time running and growing his business.
“I have years of experience working alongside other IT companies and none of them, in my opinion, have the level of understanding and knowledge that Andrew and his team can offer.”
David Short, DS Electrical Angus Ltd / Angus WISP
A technical partnership, not a takeover
David remains in control of his network. He designed it, understands it and runs it day to day. Andisa provides complementary specialist skills when expansion, resilience or strategic change needs them.
The partnership works because our backgrounds overlap. Before moving into IT, Andisa Managing Director Andrew Morrison worked in power and control electronics, designing control systems for laboratory chemical processes and industrial applications. That shapes how we design networks: rather than asking whether a connection works, we ask how the system behaves when individual components stop working, and how it recovers without someone having to intervene.
Combined with Andisa’s customer-first approach, this is one of the reasons David chose to keep working with us after using other MikroTik specialists.
Lessons for anyone planning ISP or multi-site failover
- A second line is redundancy, not resilience. Without the right architecture, it only adds bandwidth.
- Never prove a redesign on the live network. Build a replica of the core and test failover there first.
- Plan for what changes during failover. Public IP addresses will change, so build in Dynamic DNS and services that reconnect on their own.
- Segment before you scale. Separate management and customer traffic, and isolate customers from each other.
- Make remote management secure and standard, so troubleshooting never starts with a journey.
- Back up device configuration daily and automatically.
- Design for the failures you will actually get, including physical isolation of remote sections.
MikroTik failover and network resilience FAQs
Design the failover into the core routing rather than bolting it on. The core MikroTik routers must detect a lost circuit, move traffic to the second line and bring dependent services, such as PPPoE sessions, back up without anyone logging in. For DS Electrical, Andisa built and proved this across two core routers in a replica lab before it touched the live network.
The network has to be designed so PPPoE sessions can reconnect through the alternative connection automatically. On DS Electrical’s network, around 350 PPPoE customers now reconnect through the second leased line when the first fails, with no manual reconfiguration. Before the redesign, the same failure meant reprogramming hundreds of connections by hand.
When traffic moves to a different leased line, the public IP address changes. On DS Electrical’s network, each reconnection automatically forces a Dynamic DNS update, so customers and services keep using the same external DNS names. Monitoring systems reconnect and VoIP services re-authenticate on their own.
It depends on the design and how quickly a failure is detected. On DS Electrical’s network, a leased-line failure that once needed significant manual work now typically causes around five seconds of disruption before PPPoE sessions, monitoring and VoIP services are running again.
Redundancy means having spare capacity or duplicate components, such as a second leased line. Resilience is the ability to keep working and recover automatically when something fails. DS Electrical already had redundancy: two leased lines. Andisa’s redesign turned that redundancy into resilience, with automatic failover between them.
Build a replica of the core network and prove the new design there first. Andisa recreated DS Electrical’s core network at its Harrogate office, developed and tested around 6,000 lines of MikroTik RouterOS configuration, then deployed the proven configuration to the live network during a carefully chosen quiet period.
Use separate VLANs for customer traffic and network management, and isolate customers from each other. On DS Electrical’s network, this segmentation improved security and made the network easier to route and manage. It also allowed secure remote access to management interfaces without exposing them to customer traffic.
Design for failure, not just for normal operation. On DS Electrical’s network, which carries wind farm monitoring and control traffic, monitoring systems reconnect automatically after failover. In the next stage, critical monitoring services that make outbound connections will keep sending data through local Starlink backup links even if part of the backbone is cut off.
Yes, with the right design. DS Electrical is adding Starlink failover to selected sections of its backbone so an isolated section can keep authenticating customers locally and route traffic through its own backup link. Dedicated public IP addresses cannot be preserved in that emergency mode, so the design relies on outbound connections and Dynamic DNS.
Not for day-to-day running. Many owner-operated networks run well without outside help. A MikroTik consultant adds value for expansion, resilience redesigns and changes where a mistake affects hundreds of users. David Short designed and still runs his network. Andisa provides specialist skills when strategic change needs them.
Look for consultants with proven RouterOS experience on networks like yours, a safe way of testing changes, and references from live deployments. Andisa IT is a Harrogate-based managed IT provider with MikroTik expertise, chosen by DS Electrical after it had worked with several other MikroTik specialists.
Cost depends on the size of the network, how much redesign is needed, whether a test lab is required and whether you need ongoing support or a one-off project. For example, the typical cost for a day is £950.
Need MikroTik support or a more resilient network?
Whether you run an ISP, a multi-site business or a network carrying critical monitoring traffic, Andisa IT’s MikroTik consultants can design failover, segmentation and remote management that keep working when something breaks. We work alongside your team or manage the network for you, from our base in Harrogate, for organisations across Yorkshire, Scotland and the UK.
More Ways We’ve Made a Difference
Fast Flexible Communications that helped find clients
Read MoreFast Flexible Communications that helped find clients
Switching to VoIP During Office Redevelopment – YRC
Read MoreSwitching to VoIP During Office Redevelopment – YRC
Enhancing Communication at Ripon Cathedral with the 3CX Phone System
Read MoreEnhancing Communication at Ripon Cathedral with the 3CX Phone System
Automatic Leased Line Failover for a 600-Mile MikroTik Wireless ISP Network
Read MoreAutomatic Leased Line Failover for a 600-Mile MikroTik Wireless ISP Network
Andisa IT - Leading Managed Service Provider in Yorkshire, serving Harrogate, Leeds, York, and surrounding areas.
