6 ms·
Funny, my 2nd job was as employee #1 for a local ISP. We had a very similar arrangement (USR modems and all) for dial-in until we financed out and bought proper
by bane 12y ago
Funny, my 2nd job was as employee #1 for a local ISP. We had a very similar arrangement (USR modems and all) for dial-in until we financed out and bought proper hardware, I wish I could remember the details. We hosted something like 3000 users, all their websites, shell access, Usenet, everything off of just 2 or 3 Pentiums with even fewer specs than what you describe. Everything ran quite well. We even ran a couple very high volume corporate websites for some rather famous television shows of the time off of this arrangement.
I'd point out that it appears even a high volume site like HN hasn't knocked this ancient system offline.
It drives me nuts when I read about figuring out how to handle some couple hundred requests a second on racks full of modern hardware with Gigabytes of RAM. Any clearance bin laptop you can buy at Costco should be able to handle thousands to tens-of-thousands of requests per second. We're clearly doing something wrong these days.
- jason_slack 12y agoI owned an ISP in upstate NY for a few years. Basically hosting websites for clients. I bought a used APC NetShelter cabinet and put 2 large UPS's in the bottom along with and Air conditioner cooling fan in the top. I have a Sun Fire V 1 u server that was running apache. I have a few Sun Blade 100 Workstations handling mail and a few more handing DNS, NTP, etc. All I remember spending at least 25K on all of this. One interesting thing that I remember was trying to build a drive array back then. Trying to figure out how many IDE drives I could stuff into a 4U case to use as what we call NAS now a days..... I did this from my house and had to upgrade to 200amp electrical service and have dedicated circuits run to my house. In order to do this the phone company had to dig up part of the road and it took them a week.. Not convenient for all that traffic down my street.
- vidarh 12y agoThe phone company gave up on the idea of digging in our case. We had stupidly not thought the phone situation through, so we got offices first and investigated the phone line availability second. Luckily this was before de-regulation so the incumbent telco was legally obliged to provide service no matter what at a fixed price per line. Unluckily we'd gotten offices in the embassy area in Oslo. Unlucky because due to the embassies, this area was one of the first in Norway to get upgraded to an automated exchange, and said exchange had last seen a major upgrade in the 1930's or thereabout (given that this was before commonplace internet access spurred a flurry of exchange upgrades), and was near capacity - until we arrived this was not seen as a problem, as the area was fully built up and there was seemingly no reason to expect any big increase in number of lines that couldn't be accomplished just fine mainly by line splitting (multiplexing two lines onto one copper connection to the exchange) that worked fine for voice calls, but not so well for modems... Their solution turned out to be to run a massive cable largely hanging in the trees along a major road from the nearest newer exchange with decent capacity a few hundred meters away, through a hole cut in one of our windows, to a multiplexer cabinet they installed in our office. They did tell us they were going to get it sorted out "properly" but by the time we moved to new offices they still hadn't been able to. At least they did it quickly.
- jason_slack 12y agowow, running a cable through trees into a hole in a window. That sounds crazy, but a conversation piece for sure! They ran my circuit to the outside of my house, but would not drill a hold for me to connect it to the inside wiring I had already in place. I tried to do this myself, measured wrong and ended up tapping into my main power wire with a 1/2 inch 2 foot long drill bit and setting my house on fire. Thankfully the damage was minimal and I was up and running a few days later once power was restored to my house.
- vidarh 12y agoThe thing is, so much was static. It's easy to shuffle bytes. There are a few problems: People don't know or care about profiling the basic stuff like context switches and data copies; people don't have a baseline idea of what should be possible; abstractions upon abstractions upon abstractions even when it complicates the code. One of my biggest pet peeves in that respect is that people tend to not even know how many objects they create in many dynamic languages. Many people don't even know of any simple ways of finding out. My most successful application speedup (EDIT: measured in speedup per hour expended - plenty of examples of much greater speedups, but they are rarely as quick) was spending an afternoon eliminating string copies in a late 90's CMS written in C++, and cutting page generation time by 30% in a system that was already heavily optimized (it provided a statically typed C++ like scripting language that was bytecode compiled at a time when most competing solutions were using horribly inefficient interpreters, and were often themselves written in slow intepreted languages). My second biggest pet peeve is when people think system calls are cheap because they look like function calls. Sometimes you can increase throughput by an order of magnitude just by eliminating small read()'s (non-blocking filling of userspace buffer instead, and "read" from that). When I see slow throughput, the first thing I do is break out "strace" to look for unnecessary system calls... But these problems are often a direct result of abstractions that means huge numbers of developers don't know or understand the cost of a lot of the functionality they depend on any more, and can get away with that "too often" because they work on hardware that is so fast it is extremely forgiving (e.g. my new "weird unknown Chinese brand" 8-core phone does about 10 times as many instructions per second as my ISPs total computing capacity in 1995, and it's nowhere near the peak performance of todays flagship phones), ignoring the fact that at "web scale", things are not as forgiving any more: server costs quickly start adding up. But of course, most of us most of the time work on stuff that needs to handle little enough traffic that paying an extra $20k for hardware is cheaper than spending the developer time to optimize this stuff. Ironically this makes the optimization more expensive: Fewer developers ever get to obtain the experience to optimize this stuff at scale.
- bane 12y ago>My most successful application speedup ever... > people tend to not even know how many objects they create in many dynamic languages One of my biggest speedup success stories was an internal inventory reporting tool for a large-ish online retailer. The code was written in Perl, with huge amounts of data aggregation happening in memory and using loads of Perl hashes pointing to other Perl hashes point to others (and turtles all the way down). On small subsets of the data, it would run for 4 or 5 hours then spit out some graphs -- this was acceptable performance for daily reports. But as the site grew, it started taking 8 hours, then 10 hours and then finally crashing when the system simply ran out of memory. So I did a simple conversion, all the hashes pointing to hashes etc. into arrays pointing to arrays. Pretty simple in the code, only a few lines of code here and there and a couple functions to map hash keys into array indexes. The next day the entire inventory system did a run (with this one change) and finished in 10 minutes and AFAIK they still use that same system even though the site and user activity has grown a hundredfold. The developer (self-taught) of the original code asked why this worked. I walked him through the big-O complexity of arrays vs. hashes, and more importantly the amount of stuff the hash type was dragging around through memory. There used to be a fantastic site with visualizations of all the perl datatypes and we sat down and started counting bytes until he understood the difference.