Hacker News Reader

Comment by walrus01 | Hacker News Reader

As a network engineer, it's amazing the amount of work and money that it costs to do high capacity long distance singlemode fiber between major metro areas. Doing something like a 288 strand cable from Boise to Salt Lake City along railway and highway right-of-ways is a huge project that involves massive, heavy equipment, tons of people, GIS experts, civil engineers, dozens of county governments, state governments, road closures, electrical system coordination, property leases for fiber huts, diesel generators, bucket trucks, bulldozers, excavators, trenching machines, flatbed trucks, spool trucks, traffic flaggers...

I could write a 5 page list with just individual single line item bullet points of all the factors/sub-contractors/sub-systems to take into consideration for a successful inter city fiber project.

And that is just at OSI layer 1 before you even start lighting the fiber.

View on HN

I'm in the Waterloo region, which is partway through a multi-year light rail construction project. The line is really relatively simple and not all that long (19km), but took over a decade of fighting to get approved, costs almost $1B, and has been a huge undertaking to build (not to mention all the nonsense from Bombardier, who are supposed to be building the trains).

But it's been fascinating to compare this to how inner-city rail projects went in the early 20th century— basically some company would get permission, lay down a bunch of track in the middle of the street, and be running trains within a few months. You see similar stuff with early electrification projects, where basically whoever could just come along and string their wires between lamp posts.

I totally see that that's insane, and we can't have a wild west with our infrastructure, but I do also wonder if there's not a middle ground that would allow somewhat more agility with these things.

As things get more rural, and literally more wild west, it's possible to bring down the cost of OSP fiber infrastructure significantly. There's ISPs in rural western states building GPON networks along roadways/right of way (trenched/duct and aerial) with crews that consist of basically two guys and a bucket truck with a spool on it, and a $3000 fusion splicer from china.

Taobao suggests $1000-1500 is enough for a fusion splicer. https://s.taobao.com/search?q=熔接机

yes, for a basic one, there's lots of good fusion splicers for $1500-2000 from china. keeping in mind you'll need an additional $500-600 of precision hand tools, tools for dealing with OSP fiber cables, cleaning supplies and stuff to go with it, a fusion splicer by itself is not very useful.

I would perma-scroll a website with these sorts of lists.

Would you be willing to elaborate a little on your career path? Sounds really interesting.

IMHO the best way to become "the person with enable on the core routers of a noteworthy ASN" is to start from tier 1 ISP help desk and work your way upwards, while studying stuff like optics, starting with CCNA level IP networking and incrementally getting into more complicated things, and focusing exclusively on *BSD/Linux for all server platforms while completely disregarding the existence of Windows.

For people working their way up through the ranks at an ISP, it is really helpful to spend as much time cross training and working with different groups of people, responsible for different things. Spend enough time in massive-scale hosting/colocation/datacenter operations that you're familiar with how to properly set up a facility for thousands of 1U size servers, but also spend time with OSP fiber people, learn the fundamentals of high capacity PTP microwave links. Spend time with the sysadmins who run back end infrastructure like large scale DNS, SMTP and IMAP servers, billing systems. And of course, all the time, continue to increase your IP/layer 3 networking knowledge until you're comfortable on something like an ASR9010 or MX960.

Reply on news.ycombinator.com