stryk
10-06-2015, 03:40 AM
Is anyone else getting just crap download speeds from Highwinds Usenet providers?
Normally Highwinds will max out my 150mbit day & night without breaking a sweat. Very rarely there are routing issues but they always seem to work themselves out after a few hours max. However, this past week, I have gone from getting 20MB/sec to capping out at 4-8MB/sec. I have spent some time trying to figure this out. I created a free trial account with another server besides the one I have been using for a while now, since I wanted to see if it might have been limited to just one server, one datacenter, one rack, etc. Nope, it was capping out way low with both providers that I was using.
There are periods of time where it will jump back up to normal speeds. yesterday afternoon & last night, for example, I was getting the normal 130-155mbit throughput from them, but now it's back down to 4-8MB/sec max. I've tried using SSL and non-SSL connections. I tried using both providers' European servers, but only got a moderate difference in throughput.
I'm curious, is there some major routing outage going on, or some shit happening with major backbone providers (on the level of Cogent, Level3, Akamai, etc.) that I'm just not aware of? I've run every network test I can think of (that is easily accessible). Traceroutes to both providers don't return anything too out-of-the ordinary looking. There's about a 30ms latency spike once the hops reach Highwinds' gateways, but that's not that much. If there is a major outage somewhere I figure it'd be higher than that. And interestingly -- it's easy to see that 2 separate usenet "providers" are just part of some bigger company if you run tracert's on both of their nntp servers and the hops for both eventually run along the same path and end up at almost the same IP, with only the last set of IPv4 digits being different -- ie; one will end up at xxx.xxx.xxx.78 and the other will be the exact same only xxx.124 at the end instead of xxx.78.
This seems to be limited to whatever section of Highwinds' network they dedicate to Usenet servers, because Steam uses Highwinds' as a CDN -- and I can still pull my max 155mbit down from Steam without a problem all week since I've been testing this. (for those who don't know what Steam is, it's a PC/Mac/Linux gaming platform/service/store -- the biggest one by far. they've even expanded to movies and software now as well, it's a massive service that has to be making a shit-ton of money) Also, I've run speed/jitter/load-balance tests at all the normal contenders and nothing is out-of-the-ordinary there. (Speedtest.net which is Flash-based, speedof.me for HTML5, DSLReports.com which is javascript I think, etc.)
[Yes, I tried e-mailing the cust. support of the main usenet provider that I've been using, but I didn't get anything more than the boilerplate response of "We're not seeing anything systemic on our end that would indicate why your bandwidth is limited." This didn't surprise me. Also this is something cool I found while researching options for this issue, check this out: http://lookingglass.hwng.net/index.pl ("hwng" = Highwinds Network Group) This seems to be an outward-facing, publicly accessible terminal interface to a switch inside Highwinds' infrastucture, albeit with extremely limited abilities, only 5 commands. It's likely the Perl scripts are just mirroring the physical switch's outputs, but it's still fun to play around with if you wanna see how big networking works. Look at your tracert paths and, depending on where you are at, you'll see R1.DC, R2.DC, R1.AT, etc. in the DNS names of the path your packets take on their way back & forth. Those are the different routers inside Highwinds' network. I thought that was pretty neat, but I've always been a bit of a networking geek and armchair amateur engineer ;) ]
Sorry for all the nerdy networking mumbo-jumbo, but I'm running out of ideas :lol:
EDIT: Holy shit, I tried doing some more tests after posting this, and now I'm getting 446KB/sec... that's [U]kilo-bytes/sec.. wow.. yea, something is definitely going on.
Normally Highwinds will max out my 150mbit day & night without breaking a sweat. Very rarely there are routing issues but they always seem to work themselves out after a few hours max. However, this past week, I have gone from getting 20MB/sec to capping out at 4-8MB/sec. I have spent some time trying to figure this out. I created a free trial account with another server besides the one I have been using for a while now, since I wanted to see if it might have been limited to just one server, one datacenter, one rack, etc. Nope, it was capping out way low with both providers that I was using.
There are periods of time where it will jump back up to normal speeds. yesterday afternoon & last night, for example, I was getting the normal 130-155mbit throughput from them, but now it's back down to 4-8MB/sec max. I've tried using SSL and non-SSL connections. I tried using both providers' European servers, but only got a moderate difference in throughput.
I'm curious, is there some major routing outage going on, or some shit happening with major backbone providers (on the level of Cogent, Level3, Akamai, etc.) that I'm just not aware of? I've run every network test I can think of (that is easily accessible). Traceroutes to both providers don't return anything too out-of-the ordinary looking. There's about a 30ms latency spike once the hops reach Highwinds' gateways, but that's not that much. If there is a major outage somewhere I figure it'd be higher than that. And interestingly -- it's easy to see that 2 separate usenet "providers" are just part of some bigger company if you run tracert's on both of their nntp servers and the hops for both eventually run along the same path and end up at almost the same IP, with only the last set of IPv4 digits being different -- ie; one will end up at xxx.xxx.xxx.78 and the other will be the exact same only xxx.124 at the end instead of xxx.78.
This seems to be limited to whatever section of Highwinds' network they dedicate to Usenet servers, because Steam uses Highwinds' as a CDN -- and I can still pull my max 155mbit down from Steam without a problem all week since I've been testing this. (for those who don't know what Steam is, it's a PC/Mac/Linux gaming platform/service/store -- the biggest one by far. they've even expanded to movies and software now as well, it's a massive service that has to be making a shit-ton of money) Also, I've run speed/jitter/load-balance tests at all the normal contenders and nothing is out-of-the-ordinary there. (Speedtest.net which is Flash-based, speedof.me for HTML5, DSLReports.com which is javascript I think, etc.)
[Yes, I tried e-mailing the cust. support of the main usenet provider that I've been using, but I didn't get anything more than the boilerplate response of "We're not seeing anything systemic on our end that would indicate why your bandwidth is limited." This didn't surprise me. Also this is something cool I found while researching options for this issue, check this out: http://lookingglass.hwng.net/index.pl ("hwng" = Highwinds Network Group) This seems to be an outward-facing, publicly accessible terminal interface to a switch inside Highwinds' infrastucture, albeit with extremely limited abilities, only 5 commands. It's likely the Perl scripts are just mirroring the physical switch's outputs, but it's still fun to play around with if you wanna see how big networking works. Look at your tracert paths and, depending on where you are at, you'll see R1.DC, R2.DC, R1.AT, etc. in the DNS names of the path your packets take on their way back & forth. Those are the different routers inside Highwinds' network. I thought that was pretty neat, but I've always been a bit of a networking geek and armchair amateur engineer ;) ]
Sorry for all the nerdy networking mumbo-jumbo, but I'm running out of ideas :lol:
EDIT: Holy shit, I tried doing some more tests after posting this, and now I'm getting 446KB/sec... that's [U]kilo-bytes/sec.. wow.. yea, something is definitely going on.