Back in August 2016 I flew out to Zambia with four others to volunteer at the Saint Francis’ Hospital, Katete to upgrade their network.
The hospital is located in in a relatively remote part of eastern Zambia, around an hours drive from the Mozambique and Malawi border. Due to this remote location, it can be a tough place to work for staff. On our first day we had no water and experienced a 12 hour power cut (which the staff told us is a regular occurrence). UK Doctors usually go over to the hospital to volunteer for 6 or 12 months at a time.
Whilst the hospital already had a network, this did not reach into every building and it was using cat5e cabling that is now reaching the end of its life, despite previous upgrades. It was also not being used by many departments, mainly because of the poor quality of the network and the lack of facilities. The main users of the network were the accounts department, who were using an old file server donated by Bytemark in 2015.
Our aim was to upgrade the network for future use which would allow the use of more network based services. To achieve this we aimed to install 4 core network locations that would be connected using fibre running at gigabit speeds. These core locations would then connect to edge switches in each building. Due to the daily power cuts, we also planned to install a UPS in each core network rack.
We started out on day one with the intention of getting all of the fibre run within the first 2 days. Unfortunately this took longer than we planned, taking 3 days in the end. Getting the fibre run was the most important task, as without this we would be unable to connect the network together.
Tom Cairns, who was leading the group, originally went out to the hospital in May 2016 to explain to the staff what we would be doing and what work would need doing before we arrived.
Our original estimate was made based on this plan, however unfortunately when we arrived we found that this hadn’t been fully completed, meaning we now needed to change the routing of the fibre.
The first of the fibres was considered the most important in the core network, this meant it took even longer to run it, as it all had to be placed in protective conduit (which we had to install before we could run fibre).
Another issue that halted progress was the discovery on our first day that whilst some trenches had been dug for fibre, they weren’t dug under the walls and there was no conduit to run the fibre in.
This delayed the install of one of the fibres whilst we arranged for someone to dig the trenches under the walls and we acquired some conduit (which ended up being water pipe, you have to make do with what’s available!).
Once we had acquired this we started running the fibre through the trenches to begin with, then under a covered walk way the rest of the way. By the end of the day we had got it run 3/4 of the way, but we still had a building to get through the next day.
The next day we continued running the fibre, but hit our next obstacle, which was how to run the fibre through a building. The existing cat5e was run at head height then across the corridor slightly higher (with no support).
Fibre is very delicate compared to copper, so we couldn’t use the same route. After some time and discussions with hospital staff it was decided that the best route would be to run the pipe from outside of the covered walkway, and up into the roof of the building, then bring it out at the other side and back under a covered walkway to get to the next building.
At the next and final building we (with the help of hospital maintenance staff) took the cable into the roof and ran it through the roof space, until it had to go across a corridor. The maintenance staff went away and came back with a girder that had space at the top to hold and support the fibre as it crossed and went into the IT office.
On the third day I helped to run one of the other fibres within the core network, this was just run under the covered walk ways and cable tied to keep it in place. Unfortunately when we came to connect it one of the fibres had broken. I guess the lack of protection and us wanting to get the job done quickly had led to some rough handling of the fibre. Thankfully there was another pair that we could use!
Once the fibre had been run, we then set about arranging for the hospital maintenance staff to run outdoor cat6 between buildings. This was to supplement the fibre core, as well as provide access to switches in individual buildings. At the same time, we measured and ran the cables within the buildings which would be used to connect the end user devices. I spent most of the afternoon configuring some of Cisco access switches, so that we could begin to connect end user devices.
This was a Saturday so we just worked in the morning. I spent the morning in the hospital’s accounts department punching down network cables that needed to be run from the cabinet to the workstations. Whilst doing this I was able to teach the hospital’s IT technician, Shumba, how to terminate cables using keystones. This allowed him to gain a new skill and allowed us to finish the task faster.
Shumba then spent the Saturday afternoon terminating most of the remaining cables.
Unfortunately we didn’t manage to complete all of the internal cabling, as we wanted to get the complex jobs done before we left (including a core network running over fibre, and a hospital wide WiFi network).
Day 5 was a Sunday so we only worked for half of the day. When we returned in the afternoon we continued with Saturday’s tasks – running and terminating the cat6 from the edge switches. In some cases the hospital maintenance staff had run the cables so we were able to test them.
On day 6 I spent the morning configuring various edge switches (Cisco Catalyst) in the hospital so that we could begin to move existing devices over to the new network. In the afternoon I went to the school to finish terminating cables that others in the group had started.
Whilst terminating the cables I was also able to teach the 2 members of school staff how to terminate and punch down cables. We left them with a punchdown tool and some spare keystone jacks so they were able to finish terminating the cables. It allowed us to focus on other jobs which could not be done once we had left.
I returned to the school in the afternoon to configure the classroom switch whilst someone else in the group configured the firewall and the connection back to the hospital using a Ubiquti light beam. This worked fine, once I’d stopped trying to telnet to myself (oops!) and helped me improve my Cisco skills as I was left to my own devices to configure this switch.
We had 7 days planned at the hospital, once we had completed these we had two other activities planned. One of these was to go on safari in the South Luangwa National park and two nights staying next to Victoria Falls in Livingstone (with a guided tour of both the Zambian and Zimbabwean sides of the falls on one of the days).
We set out with a lot of things to do and we managed to complete the vast majority of these in the time available. This included installing a gigabit core network with a copper backup, a WiFi network throughout the hospital and a UPS backed core. We didn’t manage to install some desktop PCs or a new server (donated by Bytemark), but these were low priority jobs and could be completed once we had left.
Since returning from the hospital I’ve discovered that the important fibre that we ran has been cut in two by a rotting tree falling over in very high winds. However, the good news is our work has meant that the core has continued working, albeit with less redundancy.
Whilst this diary only summarises the work that I took part in, others in the group completed other tasks such as installing and configuring UPSs in core network racks (due to sporadic power outages), WiFi across the hospital and in the mess (where the doctors eat on an evening) and a wireless bridge to allow the mess and school to be connected up. I would have liked to have spent time working on the WiFi setup, however this was only a small task and only required one person.
Core: HP Procurve 28010-24g
Edge: A mixture of Cisco catalyst switches and some Netgear Managed switches
Sonicwall NSA 2400
Opengear CM4116 console server
Ubiquiti Unifi Access points
Ubiquiti Cloud key
Ubiqutii Unifi Light beam
Sonicwall NSA 2400
Cisco Catalyst 2950
Ubiqutii Unifi Light beam