A Network Implementation Gone South


We were at the outset of entering the market, providing IT consulting and system integration services, when we were approached by this start-up incubator, claiming that they were having issues with internet connectivity at their premises, and that they had future plans which they wanted to be ready for when the time comes.

This start-up incubator hosts a wide array of workshops and events in support of the local entrepreneurship efforts. Upon first being exposed to the institution’s network topology, how they got there, and where they planned to go, we definitely had our work cut out for us. The client, at the time, had a number of internet connections, from a number of sources, each of varying throughputs, latency, and overall performance (sounds promising already, right???).

Unfortunately though, and as is the case with most clients today (and probably always) funds were limited, so we had to make do with as little expenditure as possible. So we began to scavenge the premise for whatever hardware we could get our hands on. We finally came out with a single half-decent PC, and a wired network infrastructure that seemed sufficient for the time-being.

The strategy we decided to take, given our limited resources, was to re-purpose this PC, throw in an Intel multi-NIC, and set it up with an open source gateway solution that is built to manage at least medium-sized deployments. Our choice at the time, which seemed the most viable and reliable, with a decent community to support us, was pfSense. I really hoped that we could get dedicated and purpose-built hardware to run this role in the network, but that was out of the allotted budget, so we worked with what we had. After that we also, agreed to go for a Cisco switch, but this had to be dumbed down to an SF100 series switch, which doesn’t support VLANs (can you see the horror???).

Within a month the system was installed and up and running and all I can say is oh the troubles that came rolling in with it. We faced all types of issues, from disconnecting wireless clients, to an access point that kept on resetting itself, as well really weird connectivity issues (connections that would just cease to work), and timing-out connections that were killing internet access.

The moral of this project for me — and hopefully it will provide some insight for you in future projects — is a statement that I work by religiously today, and it is also why all my recent and current clients are a lot more satisfied with my service, and even why support calls have virtually ceased to exist… the statement is:

Never Go Cheap!

I am not necessarily speaking about the actual monetary payment, but what I mean is don’t be cheap by trying to recycle old hardware, buy something new instead; recycling may “cost less” at first, but it will only come back to bite you later pretty soon. Don’t be cheap in thinking: “oh they don’t need all that hardware, or all those features”. And finally don’t be naïve in thinking that smaller networks mean less requirements. The baseline for acceptable network performance today is relatively high, and it increases day in and day out. So believe it or not, if there’s a chance that they may need a certain feature or specification today or tomorrow, then they most probably will definitely need it the day after.

Advertisements

What do you think?

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s