Hi Dukascopy community:

If we use JForex client the best way to see broker latency is in Java console.
If you don't know how to do that, i advise you to read this article here

When you open JForex you can get the following window where you can see the latency:



In metatrader there are available some EA's to show this information.

IMPORTANT: This don't show us the order execution time !!! Just the time the conected servers take to answer us !!! The less the value of ping we get, better the chance to have better execution but one thing don't implies the other. [EDIT] We also must know this latency can change over the time depending on network conditions, server overload and other factors.

I hope this helps if want to see this information

Trade well and prospers in your way

JL

@SmithJr as i can't post this comment on his article:

Yes, i know what you say !!! But if you take more attention, the login server (https or default port 443 if you want) don't is the same server where you get the data feed (and the trace can be very different for each server as for example Dukascopy have different servers connected to different public networks), so you get the WRONG RESULTS. Other thing you must consider in your tracert or traceroute information, the "path" of the ping can be very different, and pinging the last "node" of tracert don't say you anything useful. The "tracert" or "traceroute" command only sends a packet to the provided ip or address and see where this packet goes over the network and show us the delay on each "node" (this node can be a router, gateway, lower layer switch or a forward server and others). If it stops sending us information until we get the provided ip or address, we don't know where this packet goes after. I talk about the port 10443 to determine the ip address of the correct server on netstat screen, not to do a "ping 10443" or "ping xxx.xxx.xxx.xxx:10443" or whatever. So the port i refer it's very important to get the correct ip address of the server. Other thing we must have in consideration is different brokers can use different ports for the data feed. Dukascopy use 10443 port.

But if you want to use your way, feel free to do so, but i assure you, you get the WRONG RESULTS this way.

Best regards

JL

Dear jlongo,

First of all, thank you very much for your comments! I am sure that they will be useful for all the readers.

For the interesting readers, I would like to clarify why I said in the comments of my article (sorry I don't link it here, I don't want to interfere with the Contest Rules). The ping command does not manipulate any port information, and so, you can not ping a specific port.

I have prepared the following image:



The image shows three layers of the architecture of network protocols. As one can see, the Application layer communicates with the Transport layer using the concept of port:

Port 23 for telnet,
Port 21 for FTP,
...

Obviously there is other necessary information passed between these two layers as the IP address, but this is no relevant for our discussion.

The ping tool does not follow the client - server architecture, so there is no application layer, and no transportation layer. The ping tool uses the ICMP protocol, which is at the IP level (what is called Internet layer). ICMP data is directly embedded into IP Datagrams.

This is the reason why you can not ping ports. Just addresses.

Hi SmithJr:

Maybe i'm blind or i don't know what i'm writting but i never writted the word "ping" in comments i think... Can you surprise us with the quote of my comments where i say to ping the ports ??? The only thing i say in my comments is you are chosing the incorrect ip (the one with https - default name for 443 port - on netstat window) using JForex or Metatrader. To get the ip of the real feed data server in use you need to find the ip with port 10443 on netstat screen (something like xxx.xxx.xxx.xxx:10443) and ping this ip, not the ip with https protocol. You can add all network information you want, this don't change the results.

Best regards

JL

Dear jlongo,

Thank you for your post.

Your first comment in the article is the following:


Reading this, one can misunderstand the message and think that the fact of using the HTTPS protocol (1), or to use the port 10443 (2) implies changes on the latency. In my comments in the article and in my posts here in the forum, I wanted to clarify that HTTPS or ports has nothing to do with latency, so people can avoid any confusion.

Of course you are right about the fact that different servers (one for login and one for data feed) will have different latencies, so the latency must be computed on these IPs, using ping, the java console, or any other approach. But again, the goal of my comment was to avoid any confusion for the general public, specially for those that are not IT specialist.

And this is fine, because you have shown us other approaches as well, so now people can try different approaches to get a better idea of the real latency. Thank you again for that.

All the best,