Mailing List Archive


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [tlug] Tokyo ISP Throttling? Why does my internet speed increase so much by setting up a proxy?



On Wed, Feb 24, 2016 at 1:21 PM, Stephen J. Turnbull <stephen@example.com> wrote:
> Raymond Wan writes:
>
>  > >>From home, 9 hops is required.  It spends some time around Hong Kong a
>  > bit, staying within servers belonging to the ISP.  Then it jumps
>  > directly to Oregon and then to California.
>  >
>  > Meanwhile, from work, 14 hops is required.  It bounces around Hong
>  > Kong a bit longer, including spending some time on a "HK University
>  > Backbone" -- I presume this is a shared backbone for all universities
>  > in the city.  Then it goes off to the UK, and then enters the USA.
>
> It's just like a local train, stops are basically more costly than
> distance.  The long distances are covered basically at the speed of
> light, maybe as low as 50ms certainly less than 100ms, and I bet
> you're looking at ping times in the 400-600ms range.


Not that bad.  From home about 250 ms; from work about 175 ms.

Thanks for the local train analogy -- that certainly helps!


>  > but maybe the different path and more hops plays a bigger part in
>  > download times?
>
> Path as such doesn't matter (the Russian Mafia isn't siphoning off
> packets to sell to Columbian druglords or anything like that), and a
> factor of 2 in distance is probably a secondary consideration in any
> case.  More hops matters for two reasons, both probabilistic.  First,
> every hop is a chance to encounter congestion and have packets
> dropped.  Second, every hop is a chance to encounter a configuration
> problem, and perhaps have packets fragmented.  More fragments (which
> at the IP level are just packets that can be dropped) means a higher
> probability of a corrupt packet that must be retransmitted, and that
> slows things down, even in a streaming protocol.
>
> The other thing is that I wouldn't be surprised if ICMP (Internet
> Control Message Protocol) packets get a higher priority and faster
> relay in many cases than TCP packets.


I see.  I didn't quite realize all this but it does make sense.  I
really should have spent more time understanding networks a bit more.

Experiencing a slow[er] network at work became a good reason to begin
to understand the results of ping -- I mean other than the basic "yes,
the computer is up; no, it is not"...


>  > The university is suppose to upgrade its network "over the next few
>  > years" so maybe we an see some improvement...but I don't think
>  > there's anything we can do to alter the path it takes to the USA...
>
> The university could buy connectivity from your home ISP. :-)
>
> Or you could borrow the connectivity you already have.  One thing I do
> a lot is use SSH tunnels from home.[1]  It's possible to use these
> tunnels (or specialized applications) to set up a personal VPN.  Then
> you can (quite safely) reach your box at home from work without
> opening it up to all kinds of evil.


I'd probably need to read up a lot more before attempting that.  Of
course, I'm more worried about getting the security right.

But I may have to consider it if one of my tasks is to look at "big
data" and the data is "too big" to transfer from my workplace's
Internet connection...

Thanks a lot for the pointers!  [To Scott as well about x2go; will
look into that, too!]

Ray


Home | Main Index | Thread Index

Home Page Mailing List Linux and Japan TLUG Members Links