Why is ping faster through VPN than without a VPN?

35

4

When I play an online game without VPN, the ping is around 120ms. But when I use VPN with a virtual location in Germany, the ping is only 60-70ms (which is amazing). When I use virtual location to Finland, it is again more than 160. The Internet is from university Greece.

Why would the ping through the VPN be faster than without?

Without VPN:

C:\Users\Dimitris>tracert -d prod.euw1.lol.riotgames.com
Tracing route to prod.euw1.lol.riotgames.com [185.40.64.65]
over a maximum of 30 hops:

  1    <1 ms    <1 ms    <1 ms  150.140.216.126
  2    <1 ms     1 ms     2 ms  150.140.208.73
  3    14 ms     4 ms     1 ms  150.140.208.65
  4    11 ms     6 ms     9 ms  150.140.128.50
  5    10 ms    11 ms     2 ms  150.140.128.106
  6     5 ms     6 ms     8 ms  62.217.97.30
  7     6 ms     6 ms     6 ms  62.217.100.63
  8    11 ms    11 ms     7 ms  83.97.88.69
  9    82 ms    85 ms    92 ms  62.40.112.215
 10    88 ms    84 ms    90 ms  62.40.98.150
 11    84 ms    87 ms    85 ms  62.40.98.112
 12    89 ms    91 ms    93 ms  62.40.98.152
 13    83 ms    82 ms    91 ms  109.105.102.98
 14    99 ms    88 ms    82 ms  109.105.102.97
 15    87 ms    87 ms    87 ms  80.249.209.203
 16   106 ms    88 ms    87 ms  80.249.211.64
 17     *        *        *     Request timed out.
 18    90 ms    96 ms    88 ms  104.160.141.103
 19    87 ms    88 ms    88 ms  104.160.141.107
 20     *        *        *     Request timed out.
 21     *        *        *     Request timed out.
 22    93 ms   106 ms    89 ms  185.40.64.65
Trace complete.



C:\Users\Dimitris>tracert prod.euw1.lol.riotgames.com
Tracing route to prod.euw1.lol.riotgames.com [185.40.64.65]
over a maximum of 30 hops:

  1    <1 ms    <1 ms    <1 ms  ita.fep.upatras.gr [150.140.216.126]
  2     1 ms     1 ms    <1 ms  150.140.208.73
  3     1 ms     1 ms     2 ms  upnet2fep.upnet.gr [150.140.208.65]
  4     2 ms     2 ms     4 ms  r29-vl1000.upnet.gr [150.140.128.50]
  5    15 ms    11 ms    16 ms  R29A-R29.upnet.gr [150.140.128.106]
  6     4 ms     4 ms     5 ms  upatras-rio-2-gw.kolettir.access-  link.grnet.gr [6
2.217.97.30]
  7     5 ms     5 ms     5 ms  eier-kolettir-AE.backbone.grnet.gr[62.217.100.63
]
  8    31 ms    24 ms    31 ms  grnet-ias-geant-gw.mx2.ath.gr.geant.net [83.97.88
.69]
  9    81 ms    82 ms    83 ms  ae1.mx1.ath.gr.geant.net [62.40.112.215]
 10    83 ms    81 ms    82 ms  ae2.mx1.mil2.it.geant.net [62.40.98.150]
 11    91 ms    83 ms    82 ms  ae6.mx1.gen.ch.geant.net [62.40.98.112]
 12   101 ms    95 ms    83 ms  ae4.mx1.par.fr.geant.net [62.40.98.152]
 13    82 ms    82 ms    85 ms  ndn-gw.mx1.lon.uk.geant.net [109.105.102.98]
 14    86 ms    92 ms    86 ms  uk-hex.nordu.net [109.105.102.97]
 15    86 ms    87 ms    88 ms  nl-sar.nordu.net [80.249.209.203]
 16     *       92 ms    87 ms  80.249.211.64
 17     *        *        *     Request timed out.
 18   114 ms    87 ms    88 ms  104.160.141.103
 19    88 ms    92 ms   103 ms  104.160.141.107
 20     *        *        *     Request timed out.
 21     *        *        *     Request timed out.
 22    88 ms    88 ms    90 ms  185.40.64.65

With VPN virtual location Germany:

C:\Users\Dimitris>tracert -d prod.euw1.lol.riotgames.com
Tracing route to prod.euw1.lol.riotgames.com [185.40.64.65]
over a maximum of 30 hops:

  1    49 ms    49 ms    49 ms  100.100.48.1
  2     *        *        *     Request timed out.
  3    59 ms     *       58 ms  95.172.68.229
  4    52 ms    48 ms    55 ms  95.172.71.149
  5    50 ms    78 ms    78 ms  95.172.67.5
  6    52 ms    64 ms    59 ms  77.67.73.165
  7    48 ms    49 ms    49 ms  89.149.186.170
  8    50 ms    53 ms    49 ms  4.68.63.57
  9    55 ms    51 ms    53 ms  4.69.154.71
 10    49 ms    54 ms    66 ms  4.69.154.71
 11     *        *        *     Request timed out.
 12     *        *        *     Request timed out.
 13     *        *        *     Request timed out.
 14     *        *        *     Request timed out.
 15    61 ms    77 ms    76 ms  104.160.141.103
 16    68 ms    75 ms    60 ms  104.160.141.107
 17     *        *        *     Request timed out.
 18     *        *        *     Request timed out.
 19    69 ms    64 ms    61 ms  185.40.64.65

Trace complete.

With VPN virtual location Finland:

C:\Users\Dimitris>tracert -d prod.euw1.lol.riotgames.com
Tracing route to prod.euw1.lol.riotgames.com [185.40.64.65]
over a maximum of 30 hops:

  1   145 ms   130 ms   130 ms  100.100.112.1
  2   123 ms   143 ms   129 ms  10.25.52.1
  3   150 ms   155 ms   140 ms  87.245.248.22
  4   167 ms   175 ms   176 ms  87.245.234.118
  5     *        *        *     Request timed out.
  6     *      165 ms     *     104.160.141.52
  7   167 ms   167 ms   151 ms  104.160.141.101
  8   161 ms   171 ms   178 ms  104.160.141.105
  9     *        *        *     Request timed out.
 10     *        *        *     Request timed out.
 11   164 ms   162 ms   152 ms  185.40.64.65

Trace complete.

Jim

Posted 2016-05-16T18:35:09.900

Reputation: 475

3because your traffic is taking a very very different path when using the VPN, and the VPN service provider has established a lower-latency path to the destination. – Frank Thomas – 2016-05-16T18:51:16.433

5My comment was far from a proper answer to this question. The original question was also far different then the current revision. Nonetheless I deleted the comment to avoid future pings – Ramhound – 2016-05-16T23:51:30.927

Can you also provide a traceroute to your VPN provider? This will allow you to tell which route the packets there are taking – Matthew Steeples – 2016-05-17T15:42:58.633

How to do that, what is my own VPN provider? – Jim – 2016-05-17T15:48:45.503

You should run the traceroute without -d. This will make it a little more clear what's going on, at least some of the time. – Michael Hampton – 2016-05-17T16:59:38.587

tracert prod.euw1.lol.riotgames.com with german VPN? – Jim – 2016-05-17T17:07:58.297

All of them.... – Michael Hampton – 2016-05-17T17:12:21.403

A tracertoute to the VPN addresses (with VPN off) can also be illuminating. In a sense, that fills in the part before you reach your VPN (the 145 ms or 49 ms to the first IP in the VPN case). – Yakk – 2016-05-18T13:28:34.670

Answers

28

Although it is an uncommon situation, it is possible. There could be a lot of answers.

It is probably about the internal structure of your network provider. What could help a lot, if you would give us a result of your tracert -d 1.2.3.4 command, where 1.2.3.4 is the IP what you pinged, and you would edit into your question.

Without that, I think the most probable reason is the network organization of your ISP.

The ISPs also need to pay for the network of different providers. In most cases, they are continously trying to find the cheapest / best providers, thus in most cases they use different routing for a near-area and for the far, remote locations. Of course, network for them is mostly much cheaply, while the network for the local area is mostly done by a regional organization established by a group of local ISPs or by the local government.

It is normal for the ISPs to pay only a much narrower network bandwidth from the costlier part, especially if you are from a small ISP or use a cheap package.

If you use the VPN, this VPN connected probably on the faster external network of your ISP, thus it were faster. And, the VPN provider uses probably a much faster network as your network provider.


After extension: checking the traceroutes, we can get very precise infos, what and where is the slow point in the system. Especially if we compare all of the 3 routes.

Check, were are growing the packet response time at most. In your case (the VPN-less case and in the case of the Finland VPN) it is between 83.97.88.69 and 62.40.112.215.

Checking them in an IP Whoise lookup service (for example: https://www.whatismyip.com/ip-whois-lookup/ ), we can see, that it is where your Australian packets are arriving England. It is probably a multiple sub-ocean communication cables, which is surely very costly for your cable network provider, so it tries to spare that.

What makes the things a little bit more funny: in case of your German VPN connection, that Australia -> England cable doesn't do anything. Probable the data connections into Germany are going on a highly different way.

From the Australian network configuration, there shouldn't be any difference.

I think, the reason is in your case, in Europe. The German VPN Prodiver paid a lot more to get a better global network bandwidth, as the network provider of your game server. This is the probable reason, why can it avoid that delay between .au and .uk .

The Finland VPN provider couldn't do that, and the probable reason in this case, that although Finland is a highly developed country in the region, it is probably not connected with such a well-developed global network as Germany (this, the ISP of the Finland VPN Provider had to buy a narrower bandwidth, and can't avoid the british delay).

peterh - Reinstate Monica

Posted 2016-05-16T18:35:09.900

Reputation: 2 043

I edited my question – Jim – 2016-05-16T19:56:27.070

2umm, his latency in all cases seems way too low for australia. I'm not seeing australia in any whois lookups either. – plugwash – 2016-05-17T06:42:55.647

@plugwash The first IP in the trace (150.140.216.126) names an ISP in Brisbane. But, what in this answer important is, the method to find the "weak points" (i.e. the largest ping reply delay increase) in the system and a reasoning about its coincidence with the network of the ISPs and the global cable network providers. – peterh - Reinstate Monica – 2016-05-17T15:16:00.243

@plugwash I don't know the Australian network very well, but I think if there isn't too many routers or other active network elements in the picture, then it can be not so far from the theoretical optimum calculated from the the light speed in glass (7 - 10 ms). – peterh - Reinstate Monica – 2016-05-17T15:17:38.307

2Whois on the first IP doesn't return anything useful, just a holding message from APNIC saying "we don't manage this IP". rdns on the first IP gives "ita.fep.upatras.gr" which strongly implies greece. – plugwash – 2016-05-17T15:21:00.463

@plugwash I am still surprised: $ whois 150.140.216.126|grep -i brisbane\r\naddress: Brisbane, Australia – peterh - Reinstate Monica – 2016-05-17T15:22:32.507

@plugwash But anyways, I need to replay: not the exact location is the important thing in this answer, but the method, how to find the weak points and what can be their reason. – peterh - Reinstate Monica – 2016-05-17T15:30:46.400

Well I can see reasons why a link from greece to germany followed by one from germany to the UK would be better than one from greece to the UK. That would be much less true for links from australia. – plugwash – 2016-05-17T15:36:03.507

@plugwash I agree on the condition that the OP is in Greece, which is on my opinion a false statement. Hopefully you also didn't forget, that not the exact location of the OP is the focal point of this answer and thus I won't need to mention this also a fourth time. – peterh - Reinstate Monica – 2016-05-17T15:39:56.243

1@peterh Try whois 150.140.216.126 | grep -i -C 10 brisbane for some context to that location. Specifically, see the remarks fields: APNIC is a Regional Internet Registry. We do not operate the referring network and is unable to investigate complaints of network abuse. For more information, see www.apnic.net/irt – a CVn – 2016-05-17T15:55:48.853

@MichaelKjörling That is ok, but I think Greece is still very far from Brisbane. An ISP in Greece should be registered by the RIPE which is in Amsterdam. – peterh - Reinstate Monica – 2016-05-17T16:16:18.947

3The IP address is registered in RIPE, and it is registered to University of Patras in Patras, Greece. If you got a record from APNIC, you should actually read the record and note what it says. In particular: Address ranges from this historical space have now been transferred to the appropriate RIR database. If your search has returned this record, it means the address range is not administered by APNIC. – Michael Hampton – 2016-05-17T17:03:47.803

150.140.216.126 is the default gateway on my computer. It is on Greece. – Jim – 2016-05-17T17:06:40.827

@MichaelHampton Ok, thanks. But it is still not clear to me, why isn't it enough well maintained. – peterh - Reinstate Monica – 2016-05-17T18:10:11.177

@dpiralis Ok. Then the reason of the uncommon ping reply times are very similar, but on an European and not on a global level: the ISP of the German VPN provider has a much better network quality. Next to that, Germany is a little bit closer to you as .uk or .fi. I will extend the answer with this soon. – peterh - Reinstate Monica – 2016-05-17T18:12:57.583

The greek university as many hops in the GEANT network indeed. – Rui F Ribeiro – 2016-06-06T12:35:57.703

7

A VPN forces traffic to take different routes.

Usually, ISPs optimise for bandwidth and cost, not for latency.

From WHOIS and reverse DNS, it looks like you are in Greece and the server is in Ireland.

First, let's take a look at the non-VPN path. The 62.40.* hops seem to have screwy reverse DNS (they are within milliseconds of each other but have reverse DNS that seems to indicate four different countries), but I would guess that the hop from 83.97.88.69 to 62.40.112.215 is actually a direct link from Greece to London.

That link seems to have a high latency, accounting for the majority of the total latency in the path. I would guess that there are two possible explanations. One is that the link may be taking a geographically indirect route such as an undersea cable through the Straits of Gibraltar. Another may be that the link is congested.

With the VPN, the traceroutes only show the hops after your traffic hits the VPN server. With the German VPN, we clearly see that traffic to the VPN server must be doing over a lower latency link than the aforementioned link to London. Most likely your Greek ISP or their upstream has a link from Greece to Germany that has a much lower latency than their link to London.

With the VPN in Finland, your traffic takes 150ms to even get to the VPN provider. I would guess your traffic is probably taking a rather indirect route to get to Finland (it wouldn't surprise me if it was going via London).

plugwash

Posted 2016-05-16T18:35:09.900

Reputation: 4 587

The 62.40.96.0/19 subnet is used for the GÉANT backbone across Europe, similar to "peering nets" in IXPs. So the devices might very well be within milliseconds but in different countries, or they could simply be at an IXP but named after the country which owns them. – user1686 – 2018-06-14T10:52:07.103

4

Because of routing

While you would think when you (A) ping someone else (B), the connection is A-B, there are always hops between you and B. Your packets might actually do A-C-D-E-F-B, and each connection has its own speed, latency and bandwidth.

When you ping via your VPN, they might have better routing (either random or because they pay for it), so it might be they route you A-X-VPN-Y-B, with better speed/latency/bandwith than your original path.

Or it might be worse. Both can happen.

Konerak

Posted 2016-05-16T18:35:09.900

Reputation: 1 825

Short, simple and enough explanatation – Luciano – 2016-05-19T17:50:52.117

-3

Well, if you are a gamer, and say you are physically located in China, but you want to connect to a US gaming server. Through a VPN, when you connect to a US server, you are much close the the actual gaming server thereby decreasing your ping time. And the connection is uninterrupted which increases your ping.

Another thing is that your ISP may wish to discourage certain destinations, so they create artificial congestion on the way there. The use of a VPN masks where you are heading, and the congestion disappears.

tkiara

Posted 2016-05-16T18:35:09.900

Reputation: 7

2This doesn't make much sense, even if the VPN is close to the game servers it would still require you to connect to that VPN first, no latency difference. – Sombrero Chicken – 2016-05-17T09:37:39.740

3I call spam... why on earth mention a VPN provider when the OP's already using one? – AStopher – 2016-05-17T14:20:07.533