Need to find latency between Singapore and London Servers

0

Hi Experts,

we have some servers hosted in Singapore and now we want to host some servers in London . now we want these 2 servers.so can you confirm what is min latency we can achieve between these 2 regions .currently we are using VPC peering and we are able to achieve 190-200 ms ..

asked a year ago470 views
1 Answer
0

We don't publish latency figures and instead recommend that customers test for themselves which you've done.

However, if we look at the raw figures:

Shortest path between Singapore and London (as the crow flies) is 10,841 kilometres. In a perfect world (where data is transmitted at the speed of light and there are no other delays) the round-trip time would be around 73 milliseconds. But the world isn't perfect and the speed of light in fibre optic is actually about two-thirds of the speed of light in a vacuum so assuming the best case there (again, no other delays) the round-trip time would be around 94 milliseconds.

But: The fibre paths are generally not the shortest path. Looking at the global fibre network map we can see that the path is not direct. A search tells me that the distance is somewhere between 17,000 kilometres (best case) to 21,000 kilometres (worst case). Let's take the best case here.

At 17,000 kilometres the round-trip time (in fibre-optic) is around 148 milliseconds. And again: This is assuming there are no other delays on the path. And again, that is best case. The path could easily be longer and therefore the latency could be higher.

Where could other delays be coming from?

  • If there is any place where the signal passes over copper cabling then the speed of the signal drops again to one-third of the speed of light.
  • Each router that the packets pass over introduces small amounts of delay - and even if each router only delays the packets by half a millisecond - that adds up. 20 routers would add 10 milliseconds of delay in one direction. Round-trip time means that delay is doubled. And it would not be difficult to have 20 routers between servers on opposite sides of the world - in fact, I think that's a low number.
  • AWS encrypts traffic on our backbone when it travels between regions. Encryption takes a small amount of time as well. That's an additional delay.
  • The network is shared so latency is going to be slightly variable.
  • There are probably other causes too...

But, as you can see - 190-200 milliseconds seems pretty reasonable.

profile pictureAWS
EXPERT
answered a year ago
profile pictureAWS
EXPERT
reviewed a year ago

You are not logged in. Log in to post an answer.

A good answer clearly answers the question and provides constructive feedback and encourages professional growth in the question asker.

Guidelines for Answering Questions