How do I access the turkey remotely? - page 4

 
xrust >>:



И подвергнуть систему риску взлома - тогда уже сразу раздавать индикатор, да и все. Желающих, и умеющих ломануть серверную часть намного больше чем знающих о существовании такого языка как MQL...

:-) I was expecting such a turn of events in the discussion

--

but then they could also hack the distributor's car and the indicator.

If he can't protect the server at an acceptable level.

then in all likelihood he can't protect his own.

- it's, like, 50/50% --

--

There's only one advantage to the way Dmitri described it.

Hosting is $5 - and a virtual machine is $50

 

No way - you'd have to break the distribution server first and then calculate the incoming channel, which complicates things...

 
xrust >>:

нифига - для этого нужно будет сначала ломануть раздающий сервер, а потом вычислять приходящий какнал, что осложняет ситуацию...

:-) for someone who can break into the server, it shouldn't be too hard to get past

--

Moreover, the address of the distributor can be identified in a different way, for example via email or Skype

pretending to be a client to connect to the system as a client

--

there's no point in breaking the host

it is enough to set up a correspondence - to plant a couple of e-mails, one with a unique trojan - etc. (we are talking about a situation where a person does not know how to protect the host)

So, he doesn't know what firewalls are (which can be open only for the client and the distributor), etc.

those who break the firewall are less likely to be able to do so - their services are more expensive.

 

How often do you think your MT4 is accessing the server?

 
Roger >>:

Как Вы думаете, как часто обращается Ваш МТ4 к серверу?

Put on any network sniffer and you will see - or log on FIREWALL

 
Roger писал(а) >>

How often do you think your MT4 is accessing the server?


Very often, but not much traffic, because the technology is different, and if we do the same thing via http, I don't know if we can push the cart out of the way.
 

Http more than once every 5 minutes is expensive, but if the package is small, that's fine too

 
xrust писал(а) >>

Http more than once every 5 minutes is expensive, but if the package is small it will do.

in fact (1 single computer) - for one resource about 15 "agents" by http (from advisers - with standard Windows dlls). they go every tick / every other tick. "about 15" - because the number of terminals has varied over the last year. no problems observed!
 
Shu >>:

по факту работы (1 отдельно взятый компьютер) - на один ресурс ходит около 15 "агентов" по http (из советников - стандартными виндовыми dll-ками). ходят каждый тик / через тик. "около 15" - потому что за последний год число терминалов варьировалось. проблем не наблюдалось!

and should not be observed.

The point is that EA will not be able to go more than 1 second for information at all. This is how MT4 <-> wininet.dll<-> server work.

So, the client will pound the server with requests every second. So what? That is the reason for server to withstand any load. Imagine Google or Vkontakte pounding.

I tested to check on 20 computers + 3 terminals running on each of them in these bindings request-response and requests when running from the tester!

And all the participants in the experiment (and the ISP too :) felt pretty good. The only thing is that the test is slow. The tick is processed once per second. But it is not such a big problem either.

Therefore such systems (in which a certain block of code is put on the Internet) are quite working.

 
As for the main topic of the topic - indicators (rather than signals, which Yuri started talking about), the indicators will be built using a similar system. Only the data will be sent to the client not for one tick, but for the entire history of bars. So say, saved values of the indicator line for the whole history.
We can do some calculations:
for example for one line of indicator 250 000 bars*8 bytes (time of bar) + 8 bytes (value of line) ~ 4 mb of information.
This is a heavy load. Therefore, we need to limit the construction to 100-1000 points of the indicator (depending on the number of lines).

But request-response method is suitable for these purposes.
Reason: