I did just that relatively recently in a house we bought. OS2 single mode duplex throughout the house, all converging to a trunk which is available in three locations for equipment. It's basically future proof, but also has its own well, things. You can't really plug into a duplex (I wish though), you have to put a small switch to it with SFP+ or 28 or whatever the speed you want. Higher speed switches are also a tad expensive. And then, there's the big one - PoE. That's why I also ran CAT6A next to each duplex to rooms and they're more or less for APs in the house. Overall it's definitely future proof and fantastic, but also a bit expensive if you wanna engage that fiber through the house. Pulling the cable itself isn't much of a cost at all and I recommend it.
If you just want to use LC connectors everywhere, they’re the same price. Plus single mode fiber has the advantage that there’s only two kinds and we really only use one of them.
The math worked out just to move to single mode a decade ago
Yes, 100G multimode transceivers are cheaper, but they don't use the same fiber.
100G on singlemode (100G-LR4 being the most common) uses the well-known two-strand ("duplex") fiber. Or you can get 100G bi-directional ("BiDi") over a single strand of singlemode (fiber-to-the-home often uses this).
100G on multimode is weird. As the name implies, one beam of light, sent down the core of a multimode fiber, results in multiple modes (search "Laser modes") being sent down the strand. As they overlap, it gets hard to get a clean signal out the other end.
To deal with this issue, 100G on multimode uses fiber cables containing multiple strands per direction of travel. MPO-8 and MPO-12 are common cables used for 100G multimode: It contains eight or twelve strands of fiber. Four strands are used to send, four to receive. And the prices for those cables are higher than standard duplex singlemode cable.
Finishing up the same thing. 2x OS2 pairs with 2x CAT-6A to each drop, all coming together to the network closet.
Gives me fiber for bandwidth and copper for PoE. Figured it was smarter to do both than compromise to either one, and surprisingly the fiber was cheaper than the copper to pull.
I recently pulled OS2 fibre throughout my aparment and it was surprisingly inexpensive.
Four-fibre cable was about US$ 1.5/m (here in Switzerland, I am sure cheaper elsewhere).
I picked ONTi JT-S508CL-8S as the main 10G fibre switch (direct from Ali, for about 100 bucks).
For wired Ethernet and PoE, I have a couple of KeepLink KP-9000-9XHPML-X switches (I paid about 60 buck for each, they seem to cost around $85 now). I find that they work well and use them for 10/100/1000/2500 GbE switching and to power various devices (other switches, U7-Pro-XGS AP, Zigbee dongle, home automation server, rack fans etc).
The main splice box was about 60 bucks, 24 pigtails included.
From memory, 10G SFP+ fibre modules were about ten bucks apieces. (DAC cables are cheaper, 10G copper transceivers are more expensive.)
Plus various paraphenalia (wall face plates, keystone modules, more pigtails etc), all of which was pretty cheap.
Note that I was able to borrow a fusion splicer from a friend. Otherwise they seem to start at around 500 bucks; buying one would have been the single biggest expense.
I also run a 25G path from one point in my flat to my ISP. The cabling is exactly the same but 25G switching equipment and optics are considerably more expensive and less available than 10G.
That, plus the existence of chipsets used in those switches, such as Realtek RTL8372/RTL8373 and RTL9303. Feature rich yet dirt cheap.
I particularly like per-pert PoE power monitoring on the RTL837x. I hooked that up to my netdata to get a full history of how much each device in my rack is drawing. Don't need a fancy PDU or separate power bricks for each device -- everything is powered by the $60 switch.
(This is a home network and I'm not at all bothered about single points of failure or lack of redundancy/failover.)
At 10g, sfp+ switches are cheaper and more available than 10gbase-T switches. Fiber transceivers seem reasonable, but I could be off; I've only looked a little at fiber prices, I already have cat5 in the wall, so I'm on DAC for nearby stuff and twisted pair for other rooms for the forseable future. I don't really need 10G, but it provides a bit of fun.
I'm not even sure how you'd define a great programmer. Like Justice Potter Stewart I sort of "know it when I see it". For example, I don't think anybody is going to put Rasmus Lerdorf on the Mount Rushmore of Great Programmers, but man alive is PHP really important and quite good, even at the time of release.
Imagine a private company invents a piece of technology soooo good that the US government has to issue a ban.
Apple's G4 was banned for export. Although it was not a direct order from US government. They fell into an outdated bracket of computing power exports limits. They sure did use it for advertising it.
Digital systems don't necessarily deteriorate immediately after the causal factors. Like technical debt, issues grow unnoticed and become visible gradually.
This is neat. To be honest, I never considered Linear as "fast". Seemed laggy as most web apps, but in contrast to JIRA it's lightspeed of course. Linear is great though, a real refreshment after JIRA torture.
As for optimistic routes and "fast" - maybe we ought to talk gmail first?
reply