@simshaun

No. All was by default. FileZilla with 2 connections and PhpED with one.
Please note that increasing number of connections for FileZilla won't get any performance improvements. It can not be scaled after CPU utilization reached 100% and it did stay at 100% for most of the time when FileZilla transferred the files.
I'd say that FileZilla does its best to make it looking like the transfer goes extremely fast. But it's only pretending.
If you think it makes sense to check FileZilla with 1 connection, I'll check.



I'm not sure that any server will do so without some special settings. At least I didn't see anything like that with the defaults.



may be. But wording I read from the other opponent suggested that he means the FTP client of PhpED.




Sorry, but it looks like you didn't read my comments or they wasn't clear to you.
If your ISP or server's ISP or any router in between limits the network (IP) throughoutput per connection, you can improve the situation by increasing number of connections. In the other cases, you can't. If you don't agree, please explain how it would be possible . If you compare performance of HDD, RAM, PCI-E, Ethernet with Internet perfomance, you'll see it clear that the Network is always the bottleneck. Perhaps, it's also possible that server is overloaded and can't accept files even with that performance, but it's an edge case and I'd not consider it at all.

On the other hand, if the Network performance is high enough, you can easily get your server on the knees with multiple connections. Check server load with sufficiently big number of files and sufficiently big number of connections.



It's clear because you're transferring through the Network, where the bottleneck is outside of your computer. It means that most of the time FTP client waits for the server to respond. That's why utilization is not that high.
With local transfers you'd get pure perfomance of FTP client itself.
Otherwise you're only fooling the routers and get more bandwidth.

(Sorry to the OP for taking this slightly off topic)

Dmitri, was this FTP to a server on the internet or a server local on your network?

And is that uploading by selecting the files in the editor and pressing CTRL+SHIFT+S / F to upload them?

Neither. As I told in the beginning the main purpose of the test was to eliminate the network as much as possible or you will test who's better "fooling" the network. That's why I ran the test under CentOS 5.1/vmware 6, read: local machine.



By right clicking on the project -> upload (Shift+Ctrl+P)

Ah right, well that's a real world scenario then.

Well, locally phpEd may be quicker, in the real world scenario of uploading a load of files onto a remote server, phpEd is mega slow.

Steve,

If PhpED is megaslow, FileZilla is nevertheless megaslower
What actually slow is your network, not PhpED.
Still it makes sense to use FileZilla for you and some other customers who can't afford good network connections. In this case it's possible to get quite noticable benefits from using free FileZilla.



Sure. Otherwise I would not post it.

I'm runnin 5meg DSL. Not the fastest by any means, but not slow.
Anyways, here is my stats:

===========================
Package: Joomla 1.5.9 Full Package
Destination: Dedicated server half-way across the country
Network: No other bandwidth is being used on either test:
Timed: with Stopwatch
PC: 2.61Ghz Dual-Core AMD Opteron, 2.5gb RAM, Windows XP


PHPEd, single connection: 1 hr 4 min 2 seconds
FileZilla, 5 simultaneous connections: 0 hr 22 min 03 seconds (also never above 2% CPU utilization).
===========================

I won't go into discussion who has bigger... er faster connection, but to replay to first post:

Let say that you have some sort of renaming routine.
This won't do much if you build your site on some sort of framework, 'cos if old index is dependent on new include that is not compatible with old version, than you have a major flaw in your site logic.

Especially if you changed your DB in the process.

So secondary web application server is a must, or you will have to drop down connection until everything is in order and sync with development.

would you please let us know the file sizes, like I did (the smallest, the biggest, total size, total number of files)?

There are:
724 folders
3916 files
Total size: 12.8MB
Largest file: unicode_data.php - 287KB
Smallest files: 0KB
3647 files 10KB and under
240 files 10KB - 50KB
11 files 50KB - 100KB
14 files 100KB - 287KB

So the average kb/s in Filezilla was:
12.8 * 1024 / 5 / (22 * 60 + 3) = 1.98kB/sec, or 1.98*5=9.9kB/sec for all 5 connections.

in PhpED it was:
12.8 * 1024 / (1*3600+4*60+2) = 3.41kB/sec

now, please tell me that the throughoutput like 2-3 kB per second is caused by poor FTP clients like the ones in FileZilla or PhpED

FTP server and network have some delay when executing/passing each command. The more commands you run the more delay in total. If you transfer 1 file of 12.8MB it will complete many times faster.

In other words, it has nothing to do with FTP clients like FileZilla or PhpED.
Both can provide files at speed of some dozen megabytes per second.
It's network and server can not accept such streams, at least through one FTP connection.

Regardless of how fast the connection is going in PhpED and FileZilla, the fact of the matter is the transfer I made in FileZilla was faster because of running 5x connections.

If I was only uploading single 12.8MB files, then sure I wouldn't care. After all, only 1 connection is going to be made for it either way.

correct, but it does not mean that the problem is in PhpED



I believe it's very important, though. With faster server & network, you could get better results even with one connection. Just my 2 cents.