How to organise a minimum ping? Trade server in London - page 3
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
You think you can't find a 100ms to 400ms version, why should you choose 1500ms or more?
PS . I have read all the author's research since the start of his work. If he wants to join the discussion he can do it here, on the forum, he has three or four nicknames ...It is possible to find one. Some time will even be from 100ms to 400ms, it will be up to the point when the client is draining, just like the other 95-99% of clients. If you are interested in how to drain faster, with minimal delays - this opportunity will certainly be found. Once you show profitability, delays will increase and will start to be measured in seconds.
See https://www.mql5.com/ru/forum/206828#comment_5397882 for more results of the measurements , and yes, in fact, that whole thread about ping and performance. https://www.mql5.com/ru/forum/206828/page2#comment_5401254 : "Interestingly, the idea that with our DCs with their execution times there is not much gain from the UPU is confirmed."
About getch. Guess he won't be joining any more http://tradetrade.ru/profile/hrenfx/created/topics/ :"August 12, 2014 Thank you all. Been coming to the difficult decision for a long time to completely end my public activity anywhere." That's a shame. Back in 2009 I was reading his stories about the way algorithmic departments of banks work. Although we cannot even dream about their spreads and work with synthetic instruments in retail Forex, but the outlook is completely different, I've never met such subtleties from anybody else. Judging by the tops he has collected for farewell the banks' methods have not changed in 5 years - all the same arbitrage on synthetic instruments.
I will add a couple of thoughts of my own about the role of ping. Briefly.
- Ping is a characteristic of the network, the time of the signal passing back and forth between network cards, the exchange of signals by the special protocol ICMP. The terminal and MT server may be missing or not running;
- I can't measure ping from terminal computer to server's network card with MT. Probably, MT server is blocking via ICMP;
- it's useless if new quote came to the network card of the computer with the terminal in 0.35-2.5 ms after sending it by the network card of the computer with MT server. The packet with the quote is placed in the buffer according to the port number on the computer, where it lies quietly and waits until the desired thread of the required application (terminal) gets CPU time and starts to receive and process it;
- and quite sad: the CPU workload is at its highest in the UPU, unlike in a home computer, the CPU workload sells and makes money there. The owner of the UPU, of course, tries to ensure that the processors are not idle. So, the waiting time for a thread to get processor time ready to execute may be equal or even greater than the quantum of time for which each thread is allocated a processor. It's 100 ms or more. Of course, the execution time logged in MT does not take into account the time spent waiting. This waiting time is also not accounted for in the logged ping value.
P.S. Pinging between NICs makes sense if the work is on real-time operating systems, not Windows, and is aimed specifically at performance.The considerations are flawed, although they make sense when buying the usual dead ISP VPS in "one virtual core, a gig of RAM and 20 gig of disk" mode.
People first buy the right to hustle among 100 copies of operating system on 8-16 cores among the same optimizers, and then stories about "Windows is not a real-time system" in the form of myths from the 1990s start.
Read the threadhttps://www.mql5.com/ru/forum/37453 along with the original article and my detailed explanations.
I can argue that the option of using cheap ISP VPS for Metatrader is bad and often does not give ping reduction for a lot of technical reasons.
Reiterating the thesis about theoretical impossibility (see comments to the link above) to get a normal response in a system running in slow motion mode, I'll also say about illiteracy of some hosters. Some of them have read silly tips on terminal updates blocking, and now they create operating system images with old versions of terminals permanently fixed, and they can't even save data locally because of defining folders locks.
In order for virtual servers to fulfil their technical obligations, you need to have one of the following:
Renat, what's your answer to regular (2 times in last month) problems with your hosting (my requests in service-desk)?
In theory, your UPU is the best for MT. In fact, I lost 2 times a lot of money because of problems with your hosting.
Renat, what's your response to the regular (2 times in the last month) problems with your hosting (my requests in service-desk)?
In theory, your UPU is the best for MT. In fact, I lost 2 times a lot of money because of problems with your hosting.
Yes, there were problems with one of the hosts. As sad as it is to admit it, it happens with hardware and servers.
Fortunately, we have already fixed and reconfigured everything. We merged two previously independent clusters into one, which increased fault tolerance and gave more freedom to migrate terminals.
Yes, there was a problem with one of the hosts. As sad as it is to admit it, it happens with hardware and servers.
Fortunately, we have already fixed and reconfigured everything. We merged two previously independent clusters into one, which increased fault tolerance and gave more freedom to migrate terminals.
I'm not a fan of sadomasochism... But, I have a desire to offer to test the endurance of the service.
Rashidovich! Allow me to connect an account for a signal, at least for a day.
I remember in spring I was politely shut down similar system referring to excessive load.
At the same time let's check whether your employees managed to find an optimal solution or cannot cope with such a task...)) )))
I'm not a fan of sadomasochism... But I would like to propose a test of the service's endurance.
Rashidovich! Allow me to connect an account for a signal, at least for a day.
I remember in spring I was politely shut down with analogous system referring to excessive load.
At the same time let's check whether your employees managed to find an optimal solution or cannot cope with such a task...)) )))
Renat Fatkhullin:
Yes, there was a problem with one of the hosts. As sad as it is to admit it, it happens with hardware and servers.
Fortunately, we have already fixed and reconfigured everything. We merged two previously independent clusters into one, which increased fault tolerance and gave more freedom to migrate terminals.
Renat, the problems are not isolated, and not with a single host.
A month ago (04.09) there were problems withWebzilla New York, last week (11.10) - with MQL5 New York 1. I have been losing real money due to problems with hosting.
The saddest thing is that any question about preventing such situations in the future or at least informing users in case of problems is answered with a "sorry, no one is immune to technical failures" or just keeps silent.
What steps have been taken to prevent the failure from happening again?
Why is there no rudimentary system of monitoring and alerting the user about the failure?
You keep talking about uselessnessof "cheap UPU where the terminal fights for resources", but in fact you yourself provide even more miserable service which cannot be relied upon.
And the situation is not new, a year ago I had about the same experience of using your hosting. Then there was no big loss only because small accounts were tested.
I would really like some constructive steps to make the service run smoothly on your side.
Hello!
Also, there is an opinion[link to product on Market removed by moderator] thatmany EAs bought on MetaQuotes VPSwork very mediocrely on the Market.
I would like to know. Is it a chronic (systemic) disease or can patients still hope for something?
Renat, the problems are not isolated and not with a single host.
A month ago (04.09) there were problems withWebzilla New York, last week (11.10) with MQL5 New York 1. I have been losing real money due to problems with hosting.
And the saddest thing is that all questions about prevention of such situations in the future, or at least informing users in case of problems service-desk replies "unfortunately, nobody is immune to technical failures" or just keeps silent.
You keep talking about uselessness of "cheap UPU, where the terminal fights for resources", but in fact you yourself provide even more cheesy service which is impossible to rely on.
And the situation is not new, a year ago I had about the same experience of using your hosting. At that time there were no big losses only because small accounts were tested.
I really want some constructive steps to ensure smooth operation of the service from your side.
The service is not lame in any way.
We work hard and improve our services. This is our business.
As an optional extra I'll give some direction for reflection: what difficulties are encountered by those who allow to run tens of thousands of other people's programs on their hardware. What do some people do when they buy a service for $10 for a load of 1000.