Real Software Forums

The forum for Real Studio and other Real Software products.
[ REAL Software Website | Board Index ]
It is currently Wed Oct 16, 2019 11:37 am
xojo

All times are UTC - 5 hours




Post new topic This topic is locked, you cannot edit posts or make further replies.  [ 33 posts ]  Go to page Previous  1, 2, 3  Next
Author Message
 Post subject: Re: ServerSocket.MaximumSocketsConnect Question
PostPosted: Fri Jun 08, 2012 7:17 pm 
Offline

Joined: Tue Mar 23, 2010 8:44 pm
Posts: 673
taylor-design wrote:

If I understand MadDoc correctly, he intends to develop a WE app that has WE pages for admin, but uses ServerSockets to handle desktop client connections. The ServerSockets should be just as fast as in a console app.

If that is what he is going to do then it's not really a WE type app in the normal sense. I Would think it should be much fasteer than the typical slow WE app but I'd reserve judgement to say it'd be identical. He's talking about having thousands of users. The WE app probably has more overhead than a naked console app. Certainly it will use more ram.
taylor-design wrote:
Properly coded WE apps are faster then PHP. C may be faster if the server task is CPU intensive, but at a higher development cost. If the server's tasks are bound by bandwidth and/or database responses, then C offers no advantage.

LOL. I guess every single WE app I've seen wasn't properly coded and that includes RS made WE apps. I understand your thinking php is interpreted and WE is compiled. WE apps are supposed to be much faster but the reality of every WE app shows me how slow they are. Search the forums and see what other real world WE app developers say and keep in mind they probably don't have any php experience.

Please show us 7 WE app live hosted examples of real world use vs. their php counter parts.
Let's not forget to compare server resource use because that is significant.
My experience has shown php is about 3 to 5 times slower than C but still faster than WE apps and using less resources.


Quote:
Not having a persistant connection requires you to either have the server contact your app when new data becomes available -HTML 5 can do this and should be able to work with a RS desktop client unless there is something not up to spec with RS.

Or, the less professional way is to have your app periodically poll the server checking if there is any new data for that client.


taylor-design wrote:
At the level MadDoc is working at, it's all the same. Server-Sent events depend on the client contacting the server first. WebSockets are an always open connection. He has the same choices before him whether he uses an HTTPSocket or a TCP socket.

Any solution that involves the server actually initiating contact with the client will require the gamer to setup port forwarding on his firewall or broadband modem.

He's talking about having thousands of connects.
It is not all the same and he wants to design it correctly to be as efficient as it can be within reason with the future possibility to scale.
My understanding of SSE is that the client does not have to continously contact the server. That is the entire point. After initial contact the server updates the client only when needed thereby reducing unwanted server overhead.
SSE can work over web ports so for example when using a web browser or any other app that can listen on port 80 no port forward may be necessary. I thought RS httpsocket could listen on port 80 (maybe authentication is required¿) but I dont remember.

Your example of buying a used server and hosting it isn't very good in my opinion. You spent $300 on a old server with no OS and very used hard drives and paying a more than twice as much as a mac mini colocation or suitable VPS. Who would trust their business with old used hard drives?

If you want to run the newest and best server software on old hardware you might suffer a performance penalty.

Hard drives fail and when they fail on colocated servers its a huge problem. You dont just walk into the next room and change them when they are colocated. You can have serious down time which may involves days of shipping or an expensive trip depending on the location.

So to be realistic let's add $250 for two new hard drives.
That's about $600 including shipping for the server and new drives still maybe no OS with older hardware speeds and paying $75 a month.

For much less money you can rent a VPS and never have to worry about the hardware because if something goes wrong the tech team will switch you over o another machine plus they often help you scale to fit your needs. It is true the VPS machine may not have the identical power of your used box but realistically hardly anyone uses 100% of their box most of the time. The VPS might also be quicker with DB access if hey raided the drives and your setup didn't because theres only two drives. Mirror raiding is for backup not speed increasesso they do not apply here.

Also compare that with buying a new mac mini at $999 for the latest greatest everyhing. Hardware, Server OS, warranties and less than half the monthly cost to colocation it $30 a month.

Your plan of paying $75 per month for old nonscalable hardware on a non mac server I dont like.

In 1 year your plan has cost you more than the difference of buying a new mac mini and hosting it and the losses in dollars continues to skyrocket every month. A difference of $45 dollars a month.


Top
 Profile  
 
 Post subject: Re: ServerSocket.MaximumSocketsConnect Question
PostPosted: Fri Jun 08, 2012 8:29 pm 
Offline

Joined: Wed Mar 22, 2006 11:15 am
Posts: 712
Location: Southern California
J.Sh3ppard wrote:
If that is what he is going to do then it's not really a WE type app in the normal sense. I Would think it should be much fasteer than the typical slow WE app but I'd reserve judgement to say it'd be identical. He's talking about having thousands of users. The WE app probably has more overhead than a naked console app. Certainly it will use more ram.


If nobody is viewing a web page then there isn't going to be any additional CPU overhead. The WE portion may incur a RAM penalty, but not much. I don't see any reason why he can't put his code in a WE app if it makes it easy to build some admin pages.

I would test this to be sure (always test), but ServerSocket performance should not be different in a WE app.

Quote:
LOL. I guess every single WE app I've seen wasn't properly coded and that includes RS made WE apps. I understand your thinking php is interpreted and WE is compiled. WE apps are supposed to be much faster but the reality of every WE app shows me how slow they are. Search the forums and see what other real world WE app developers say and keep in mind they probably don't have any php experience.


I do have PHP experience and WE is potentially faster. WE gets a bad rap on speed because of the launch time, and because people write code without thinking about the client/server round trips that are generated. There are a few other pitfalls (server configuration; IE8 and below having JavaScript engines that are really too slow for WE; sensitivity to high latency or slow Internet connections). But a properly coded WE app with the proper resources behind it is fine.

Real's own customer entry app is not slow. It's as responsive as any AJAX app I've seen. It's certainly better than constantly reloading PHP pages. Of course you can create an AJAX site that uses PHP to get data and not constantly recreate entire pages. I wouldn't expect the PHP to execute any faster than compiled Realbasic.

Quote:
Please show us 7 WE app live hosted examples of real world use vs. their php counter parts.


That would tell us nothing. Coding a WE and PHP solution for the same problem would allow a proper comparison.

Quote:
My experience has shown php is about 3 to 5 times slower than C but still faster than WE apps and using less resources.


So you can provide us with 3 URLs that point to 3 web applications performing identical tasks but coded in C, PHP, and WE that we may time them? And you've made the source code available to each?

Quote:
He's talking about having thousands of connects.
It is not all the same and he wants to design it correctly to be as efficient as it can be within reason with the future possibility to scale.


My point was that HTML5 doesn't offer anything different from his choices now. His three choices are:

1) Client polls server for information at a fixed interval, opening/closing a socket each time.

2) Client keeps an open socket.

3) Server pushes to client by opening/closing a socket when information is available. This requires the client to be publicly accessible. For the typical client that means port forwarding through a firewall, router, modem, etc.

HTML5 doesn't do anything other than describe the official way to do 2 from a browser, including new APIs in JavaScript.

Quote:
My understanding of SSE is that the client does not have to continously contact the server. That is the entire point. After initial contact the server updates the client only when needed thereby reducing unwanted server overhead.


SSE is a client initiated HTTP request that is held open so the server can keep sending data as "events" occur. It also happens to be effectively unidirectional, server to client. It's basically an HTTP request with a response body that never ends. MadDoc could connect to a SSE URL using HTTPSocket, but he would have to send the correct request, and parse the incoming response body. He would be better off just keeping an open TCP connection that is bidirectional.

His choice is which overhead he wants: the memory and occupied socket of an always open connection, or the CPU time of connections which open/close.

Let's say the OS limit on sockets is 65,000. MadDoc has 100,000 gamers that need to communicate with the server, but the communication is infrequent and small, i.e. he is limited not by bandwidth or CPU but by available sockets and RAM. In this case the choices would be 1 or 3.

Now let's say that MadDoc has a different game. There are only 500 users, but their desktop apps are constantly talking to the server. Opening/closing connections is going to increase latency and take up CPU time. Now the obvious choice is 2, just keep 500 open connections. The client and server are still only talking when there's something to talk about.

Quote:
SSE can work over web ports so for example when using a web browser or any other app that can listen on port 80 no port forward may be necessary.


Web browsers do not listen on any port. SSE is not the server opening a connection to the browser, it's the browser holding open a HTTP request to the server. Port forwarding is necessary for any machine that is behind a firewall or router if that machine is going to accept incoming connections.

_________________
Daniel L. Taylor
Custom Controls for Real Studio WE!
Visit: http://www.webcustomcontrols.com/


Top
 Profile  
 
 Post subject: Re: ServerSocket.MaximumSocketsConnect Question
PostPosted: Fri Jun 08, 2012 8:59 pm 
Offline

Joined: Wed Mar 22, 2006 11:15 am
Posts: 712
Location: Southern California
J.Sh3ppard wrote:
Your example of buying a used server and hosting it isn't very good in my opinion. You spent $300 on a old server with no OS and very used hard drives and paying a more than twice as much as a mac mini colocation or suitable VPS. Who would trust their business with old used hard drives?


* I happened to replace the drives in that case. But I wouldn't be afraid of "very used hard drives" if they passed all tests and were in a RAID array.

* Linux is free.

* Macminicolo.net starts at $35 a month for 200 GB of bandwidth. My client is getting 2.5 TB of bandwidth per month for $75. If you know of a Mac mini colo that offers 2.5 TB per month at $30 with similar speed (latency and peak speed) and data center features, I would love to know about it so I have another option to recommend in the future.

* I already stated that their VPS was not suitable even at $200 a month, the peak allocation of resources available.

Quote:
If you want to run the newest and best server software on old hardware you might suffer a performance penalty.


The box I linked to is faster then the mini's Apple was shipping less than a year ago. My client has a different box, but it's also faster. Are you saying a 2010 Mac mini is no good for hosting?

Quote:
Hard drives fail and when they fail on colocated servers its a huge problem.


Only if you don't have a RAID array and/or redundant servers. My clients always have at least one. And if the server has hot swappable bays, it's no problem at all.

Quote:
For much less money you can rent a VPS and never have to worry about the hardware because if something goes wrong the tech team will switch you over o another machine plus they often help you scale to fit your needs.


In that particular case the VPS was out of steam. There was no more scaling up.

Quote:
It is true the VPS machine may not have the identical power of your used box but realistically hardly anyone uses 100% of their box most of the time.


We are in a thread talking about high performance applications of Real Studio. I have no idea how popular MadDoc's game will prove to be. But if it proves to be popular:

* A dedicated server will last longer then a VPS before he has to scale to multiple servers and Realbasic apps.

* If his game is popular and making him money, it will probably be more cost effective to scale out with his own hardware. That's something he would have to check at the time he does it.

Don't get me wrong. I like VPS hosting for solutions that fit within their price/performance envelope, and I've pointed clients in that direction. Past a certain point in terms of capacity just hosting your own hardware makes more sense.

Quote:
The VPS might also be quicker with DB access if hey raided the drives and your setup didn't because theres only two drives.


Nope, not in this case.

Quote:
Your plan of paying $75 per month for old nonscalable hardware on a non mac server I dont like.


That's because of your false assumptions :wink:

_________________
Daniel L. Taylor
Custom Controls for Real Studio WE!
Visit: http://www.webcustomcontrols.com/


Top
 Profile  
 
 Post subject: Re: ServerSocket.MaximumSocketsConnect Question
PostPosted: Fri Jun 08, 2012 10:36 pm 
Offline

Joined: Tue Mar 23, 2010 8:44 pm
Posts: 673
This thread is taking too much of my time.



taylor-design wrote:
I don't see any reason why he can't put his code in a WE app if it makes it easy to build some admin pages.

[/quote]

Sure. In all reality his game probably won't have a billion players so go ahead Doc use WE but I wouldn't have your gamers or customers accessing a lot of WE pages.


taylor-design wrote:
Properly coded WE apps are faster then PHP.


I'd like to see your comparisons of why you think this.
Comparing CPU crunching isn't a realistic comparison. Using the web is a lot more than just CPU crunching and as I've said php is interpreted and RS apps are compiled, they are supposed to run faster. The problem is the web issue that seems to cause a huge delay.

Since WE/web is hugely about displaying UI to the client browser. Write up some php and WE apps that do the same thing such as show photos, accept uploads, etc. and let's compare them on the same server.

I had previously given up on WE but if you can show me it's as fast and as reliable as php, then I may consider using it.


taylor-design wrote:
C may be faster if the server task is CPU intensive, but at a higher development cost.


What developmental cost are you talking about?
C is free, WE is not.

For example if you're writing on a Mac server Xcode is free.


taylor-design wrote:
If the server's tasks are bound by bandwidth and/or database responses, then C offers no advantage.


This also may not be true. I recall many people saying the RS database plugins are very slow compared to other database access. If you're curious do a test with MYSQL.



taylor-design wrote:
The WE portion may incur a RAM penalty, but not much.


I guess that depends on what you're considering a lot of overhead.
A blank WE app doing nothing uses .8 CPU and 6.1 megs on my Mac.
Who knows how much ram/CPU will be used once it starts doing stuff.

If you scale this up for multiple servers you're paying more of a price.


taylor-design wrote:
* I happened to replace the drives in that case. But I wouldn't be afraid of "very used hard drives" if they passed all tests and were in a RAID array.


MD and others do not have the luxury of getting freebie hard drives so what you wrote does not apply.
Hard drives in servers are well known to die sooner than in desktop machines. It's because of all the use they get.
I'm sure you know data serving and manipulation uses a lot of read/writes on the disk shortening it's life.

And replacing drives is often not as easy as you claim when colocating. If you're colocating somewhere, you still have to get there or ship your drives and have the techs do the swap. They often charge you money to do that and the process can take a few days (downtime) from the shipping and that's assuming you've got backups ready to go and waiting.


You talk about mismatched things :

You recommend your client to buy a cheapo used server and you give him hard drives so we know he's on a budget or is a cheapskate but then you try to explain your excessively high monthly hosting fee of $75.00 a month because it provides 2.5TB of bandwidth. Reality is if your client is so cheap to be buying a used noname server using freebie hard drives he isn't using anywhere near 2.5TB's of data a month so he's over paying by a lot.


How much monthly bandwidth does your client actually use with his used cheap server, lol?
In one year he's going to be paying 3 times the money in hosting than his server cost.



Somewhere in the thread I think you asked about VPS hosting plans that offer 2.5 TB's for less than $75.00.

Well starting at $15.00 Dreamhost.com has a VPS plan offering unlimited bandwidth/storage.
Don't know how realistic that is or if they cap it but for some of your clients you may want to let them know.

http://dreamhost.com/servers/vps/

taylor-design wrote:
The box I linked to is faster then the mini's Apple was shipping less than a year ago. My client has a different box, but it's also faster. Are you saying a 2010 Mac mini is no good for hosting?


Why would you bring up old Mac minis? Who cares?
My old Macs are free and they could serve too.
I'd rather have a used mac mini server with snow leopard server for $300.00 (same price as your noname) and colocate it for much much less a month.

Quote:
For much less money you can rent a VPS and never have to worry about the hardware because if something goes wrong the tech team will switch you over o another machine plus they often help you scale to fit your needs.


taylor-design wrote:
In that particular case the VPS was out of steam. There was no more scaling up.


I don't know what you're talking about here?
You can almost always scale up your VPS to another higher plan or get a dedicated server, etc.

taylor-design wrote:
We are in a thread talking about high performance applications of Real Studio. I have no idea how popular MadDoc's game will prove to be. But if it proves to be popular:

* A dedicated server will last longer then a VPS before he has to scale to multiple servers and Realbasic apps.


Used old drives and a slower CPU based server won't last as long as VPS or dedicated that you can easily switch from and upgrade.

What's easier to do: Start with a low cost VPS and scale as you need without any hardware worries or
Dealing with a dead server and a over priced monthly plan you're probably locked into.
VPS or renting a dedicated is easier/smarter and buying a new longer lasting mac mini also smarter in my opinion.

Quote:
The VPS might also be quicker with DB access if hey raided the drives and your setup didn't because theres only two drives.


taylor-design wrote:
Nope, not in this case.


You don't know that. The VPS/rented dedicated could have much faster professional grade hard drives as they often do. 15,000 rpm, 10,000 rpm, etc. THEY ARE faster than normal 7200 rpm desktop drives which is probably what you've got in that used 1u. Depending on how much you pay they may even have ram drives/SSD.


Top
 Profile  
 
 Post subject: Re: ServerSocket.MaximumSocketsConnect Question
PostPosted: Fri Jun 08, 2012 11:54 pm 
Offline

Joined: Wed Mar 22, 2006 11:15 am
Posts: 712
Location: Southern California
I'm going to reply to your statements regarding Realbasic, and separately to the rest of your post.

Quote:
The problem is the web issue that seems to cause a huge delay.


Let's pick an example everyone has access to: where is the delay in Real's demo app? What delay are you talking about?

Quote:
What developmental cost are you talking about? C is free, WE is not.


Developer time.

Quote:
This also may not be true. I recall many people saying...


Let's have less of what people say and more of what can be proven in repeatable tests.

Quote:
I guess that depends on what you're considering a lot of overhead. A blank WE app doing nothing uses .8 CPU and 6.1 megs on my Mac.


Apache idles at 25 MB and roughly the same CPU on my Mac.

Quote:
Who knows how much ram/CPU will be used once it starts doing stuff.


I do because I've actually bench marked it. A plain WE session takes about 50K of memory. Naturally this is subject to whatever the developer does with his pages. If you load and cache 1 MB of data per Session, then your Sessions will take up 1 MB each.

CPU only goes up with activity. How much depends entirely on the activity. It does not stay up. If the user is idle (i.e. reading the page) then the app is idle.

_________________
Daniel L. Taylor
Custom Controls for Real Studio WE!
Visit: http://www.webcustomcontrols.com/


Top
 Profile  
 
 Post subject: Re: ServerSocket.MaximumSocketsConnect Question
PostPosted: Sat Jun 09, 2012 12:32 am 
Offline

Joined: Wed Mar 22, 2006 11:15 am
Posts: 712
Location: Southern California
As for the rest of your post...

The reason this thread is taking too much of your time is because you insist on arguing points that are established, or that you could not possibly know (i.e. my client's needs). You are argumentative to the point of being offensive, speaking to me as if I'm clueless about servers and calling my client a "cheapskate." I'm pretty sure at this point that I've deployed and managed more server based solutions then you have, and that I'm more intimately familiar with Internet protocols and scaling client/server solutions.

Let's go point by point:

* Server drives do not die sooner then desktop drives. If anything they have a more consistent environment (no knocks, no thermal cycling, no power surges). It might surprise you to know that the chance of failure with a drive 1 year old or less (manufacturing defects) is the same as with a 7 year old drive (age). Absent defects, the average drive can expect to see 9-11 years of use, though the risk of failure does start to go up from 6 years on. That's from actual studies of drives in real world use. New or old, use a RAID and backup your data.

* Replacing a drive in a hot swap RAID server involves no downtime or difficulty.

* How old do you think the drives are in those VPS and leased servers? You can guess the age of a leased server from the CPU, and the one's I've seen were 3-5 years old.

* My client needs >500 MB of bandwidth per month, putting them in the $100 tier at a mini colo. They do not need a $1,000 Mac mini server when a $300 used Dell is just as fast. I've recommended Mac minis and macminicolo.net where it made sense. It did not this time.

* Go ahead and run a complex, SQL backed web app on a $15 a month (300 MB limit) VPS and serve 1 TB of data in a month. See how responsive it is (and how fast your account is throttled) :lol:

* Used Mac mini servers start at $600 on eBay. Without hot swap drives.

* I personally tested my client's $200 a month VPS against their server. The VPS was slower in all respects, including database queries (much slower). A VPS means you are sharing hardware with other people. Hard drives which are handling multiple web and SQL tasks do not make for fast SQL queries, and I have yet to see a VPS backed by a SSD.

* My client could have leased a server. A leased server with comparable RAM and drives would have been at least $175 per month. It also would have been about the same age.

I don't dislike VPS solutions. I often recommend them. If your needs fall within their price/performance envelope, they are great. But when your VPS bill heads north of $100, in my experience you can get a lot more performance for less money with a dedicated server. Even an "old dying cheapskate" server.

I have no idea if MadDoc is doing this as a hobby or business venture or what. I don't know his budget, nor do I know if he will have 2 players or 20,000. If I had a game that I expected to start with hundreds of users I would start with a dedicated server...the first of what would hopefully become a farm...not a VPS. And I stand by my assertion...having tested it personally...that a used server off eBay will out perform a VPS at the same monthly price point as the server cohost.

_________________
Daniel L. Taylor
Custom Controls for Real Studio WE!
Visit: http://www.webcustomcontrols.com/


Last edited by taylor-design on Sat Jun 09, 2012 4:36 am, edited 1 time in total.

Top
 Profile  
 
 Post subject: Re: ServerSocket.MaximumSocketsConnect Question
PostPosted: Sat Jun 09, 2012 2:31 am 
Offline

Joined: Sun May 20, 2012 3:47 am
Posts: 22
Quote:
Any solution that involves the server actually initiating contact with the client will require the gamer to setup port forwarding on his firewall or broadband modem


This could be automated via UPnP or with NAT traversal TCP/UDP hole punching which is exactly what i am currently developing in RB
http://www.youtube.com/watch?v=7cBPZihAsgo

You could even go as far as to develop WE/Desktop App to run alongside n2n for the network game bridging side of things which is ideal for a gaming project; also you wouldn't need to even worry about Port Forwarding etc..
http://www.ntop.org/products/n2n/

:)


Top
 Profile  
 
 Post subject: Re: ServerSocket.MaximumSocketsConnect Question
PostPosted: Sat Jun 09, 2012 2:41 am 
Offline

Joined: Sun May 20, 2012 3:47 am
Posts: 22
:cry: Oh and i forgot

Quote:
Properly coded WE apps are faster than PHP


:lol: Somebody lives in cloud cuckoo land and it certainly isn't me...


Top
 Profile  
 
 Post subject: Re: ServerSocket.MaximumSocketsConnect Question
PostPosted: Sat Jun 09, 2012 4:15 am 
Offline

Joined: Wed Mar 22, 2006 11:15 am
Posts: 712
Location: Southern California
n00bi3 wrote:
Quote:
Any solution that involves the server actually initiating contact with the client will require the gamer to setup port forwarding on his firewall or broadband modem


This could be automated via UPnP or with NAT traversal TCP/UDP hole punching which is exactly what i am currently developing in RB


There's nothing to "hole punch." His game server will be public.

Hole punching is a P2P technique that requires two clients to first negotiate through a common server, and then open a connection to each other. Aside from not always working, once the connection is closed a second connection would require a second negotiation. It doesn't allow you to arbitrarily push data server to client at any time past a NAT.

Now if his game architecture is P2P, then that is one possible solution. He specifically said it's client/server.

n00bi3 wrote:
Quote:
Properly coded WE apps are faster than PHP


:lol: Somebody lives in cloud cuckoo land and it certainly isn't me...


Post the URL to a PHP app you've written that is as responsive as Real's demo app. Don't forget to include a source code download link and full hosting specs so that it may be publicly tested, compared, and confirmed.

I'm sure we won't have to wait long :roll:

_________________
Daniel L. Taylor
Custom Controls for Real Studio WE!
Visit: http://www.webcustomcontrols.com/


Top
 Profile  
 
 Post subject: Re: ServerSocket.MaximumSocketsConnect Question
PostPosted: Sat Jun 09, 2012 5:02 am 
Offline

Joined: Sun May 20, 2012 3:47 am
Posts: 22
Quote:
Hole punching is a P2P technique that requires two clients to first negotiate through a common server


Complete nonsense please show me the MITM (Man in the middle) in our latest haskell project
http://sourceforge.net/p/portfusion/home/PortFusion/

Next you will be telling us that RBSQL is superior to MySQL

No offence intended but i think you should just stick to writing pretty little web controls

:lol: :lol: :lol: :lol:


Top
 Profile  
 
 Post subject: Re: ServerSocket.MaximumSocketsConnect Question
PostPosted: Sat Jun 09, 2012 6:42 am 
Offline

Joined: Wed Mar 22, 2006 11:15 am
Posts: 712
Location: Southern California
n00bi3 wrote:
Complete nonsense please show me the MITM (Man in the middle) in our latest haskell project
http://sourceforge.net/p/portfusion/home/PortFusion/


From the page you linked to (emphasis mine):

Quote:
1 PortFusion instance started on a publicly accessible server acts as a gateway to the public internet
2 PortFusion instances started in networks behind corporate firewalls connect to the public PortFusion instance and establish tunnels for traffic relay
clients from the public internet connect to the public PortFusion instance


Client <-> Public PortFusion instance <-> PortFusion instances behind firewalls

The middle man is right there in the middle. In the case of PortFusion it's not even TCP hole punching, it's relay via a gateway. This has nothing to do with MadDoc's question.

Quote:
No offence intended but i think you should just stick to writing pretty little web controls


I get this from someone who took 19 days (nearly 3 weeks) to figure out how ServerSockets work after a solution was explained in the first response to his question:
viewtopic.php?f=2&t=44068

Now you're lecturing the forum regarding Real Studio server and WE performance, as well as the proper way to implement a scalable, high performance game server. :roll: And intentionally insulting someone who has actually written Realbasic business servers comparable to what MadDoc wants for a game server.

No offense intended, but study the forum FAQ: faq.php#f00

_________________
Daniel L. Taylor
Custom Controls for Real Studio WE!
Visit: http://www.webcustomcontrols.com/


Top
 Profile  
 
 Post subject: Re: ServerSocket.MaximumSocketsConnect Question
PostPosted: Sat Jun 09, 2012 7:15 am 
Offline

Joined: Sun May 20, 2012 3:47 am
Posts: 22
:lol: If you had bothered to read the PortFusion Wiki you might have noticed that you only need to run a Porfusion client on host A
and a Portfusion Server listener on host B.

so 1+1=2 (so where is the middleman in which case it would be Host A, Host B, Host C) = 1+1+1=3

BTW YOU :lol: were the one that made the bold statement that a RB WE App would outperform a PHP App (More rubbish)
and rambling on about Mini Apple Macs for hosting a pontential enterprise grade gaming application.

Check you out with the Wordpress PHP @ http://webcustomcontrols.com/ :lol:

End of... Your getting kinda boring and deluded now

MadDoc maybe you check this out, I have a friend that can help if you are looking to get something up and running, personally i would ditch the RB webapp idea. RB web edition is not mature enough at the moment just go with Lua.

http://getmoai.com/


Top
 Profile  
 
 Post subject: Re: ServerSocket.MaximumSocketsConnect Question
PostPosted: Sat Jun 09, 2012 10:03 am 
Offline

Joined: Wed Mar 22, 2006 11:15 am
Posts: 712
Location: Southern California
n00bi3 wrote:
:lol: If you had bothered to read the PortFusion Wiki you might have noticed that you only need to run a Porfusion client on host A
and a Portfusion Server listener on host B.


I did read the Wiki. There is no example of PortFusion connecting two computers which are both behind NATs/firewalls, with no ability to listen to ports on a public IP, without a middle computer with a public IP. The two computer examples I read included the requirement to open the firewall for one of the computers.

And once again, this has nothing to do with MacDoc's question.

Quote:
BTW YOU :lol: were the one that made the bold statement that a RB WE App would outperform a PHP App (More rubbish)


There's nothing bold about it. In general, a compiled WE app will execute code faster than a PHP interpreter. Go ahead and benchmark it. I dare you.

Quote:
and rambling on about Mini Apple Macs for hosting a pontential enterprise grade gaming application.


Are you saying enterprise grade applications can't be hosted across a farm of Mac minis? You do realize that for most of the past decade farms at places like Google, Yahoo, and Facebook had less powerful nodes then a Mac mini, do you not? Heck, many of their nodes are probably still less powerful.

Edited: removed my last quote/response because you're just not worth it.

_________________
Daniel L. Taylor
Custom Controls for Real Studio WE!
Visit: http://www.webcustomcontrols.com/


Top
 Profile  
 
 Post subject: Re: ServerSocket.MaximumSocketsConnect Question
PostPosted: Sat Jun 09, 2012 11:18 am 
Offline

Joined: Sun May 20, 2012 3:47 am
Posts: 22
:lol: No i am saying that you are a complete fruitcake :roll: and that you somehow believe in that little mind of yours that you are some offspring guru brainchild of the Apple co-founder Steve Wozniak and somehow just so happen to know EVERYTHING about EVERYTHING and that the internet is merely a huge farm of apple mac mini's :lol: Plz get a grip

You really should see your doctor about upping your daily dose of meds..... :lol:

:lol

With all due respect FOS somehow springs to mind


Top
 Profile  
 
 Post subject: Re: ServerSocket.MaximumSocketsConnect Question
PostPosted: Sat Jun 09, 2012 11:35 am 
Offline

Joined: Sun May 20, 2012 3:47 am
Posts: 22
Quote:
Now you're lecturing the forum regarding Real Studio server and WE performance, as well as the proper way to implement a scalable, high performance game server. :roll: And intentionally insulting someone who has actually written Realbasic business servers comparable to what MadDoc wants for a game server.


Please provide us with a link to an online web application that YOU have developed that can be comparedd to what MadDoc is wanting for a Game Server.? :lol: so that i can flood it and watch it CRASH...


Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic This topic is locked, you cannot edit posts or make further replies.  [ 33 posts ]  Go to page Previous  1, 2, 3  Next

All times are UTC - 5 hours


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum

Search for:
Jump to:  
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group