Takes me forever to load up anything here and I have cable modem. No other site is like that... Anyone know why?

 

I don't think it is our server, it seems to be humming alone nicely.

Check out the 2nd post in this thread to determine if it is a "hop" between you and us...

https://www.f150online.com/forums/sh...hreadid=143040

 

Ok...thanks....

Did it and got 12 hops....

 

If I may...

'Hops' have nearly nothing to do with overall performance of a network/website/whatever. A hop is simply a device that the packet must travel through to reach its destination.

For Example, in order for me to access this wonderful site (kiss kiss), my packets (data requests & data coming back) must first go to a northern Detroit suburb, over to Chicago, Down to St. Louis, then over to Atlanta where it bounces around a bit until it hits the hosting company that the server resides with. My data packets pass through (or 'hop') through 13 different devices. Now, the real fun part comes in when you look at the actual response time of the network. My packets SCREAM through those 13 hops in an average time of 44ms (miliseconds). I can't even blink my eye that fast. A blink of your eye takes a staggering 100 milliseconds!

Now, Joe blow could have a mear 5 hops but on a congested network, thus resulting in delay of 80-100ms or more!!

A simple 'ping' will show you the latency of the network. However, even this isn't a TRUE test of speed. Pings and trace routes only use tiny ICMP packets to do the work, things can be different when your moving bigger hunks of data around, as all those 'hops' may introduce some more latency while trying to figure out how/when/where to route the massive amounts of data flowing through them.

Now that your head is spinning, I feel the site has issues as well. I have many random LONG delays throughout the day. It could be the server, the network, the hosting providers, etc... Only some nice long network/host analysis would tell the real tale...

Food for thought, oh great webmaster!

 

One thing that impacts peoples perception of performance is large images.

The image in your signature is a great example. its 41k and 600 pixles ... That completely fills the screen for a user that uses either 800x600 or even 1024x768 will have to scroll once you allow for the space around the forum.

Everything adds up

Doug

 

Doubtfull and entire page of 41k 640x480 pics would even phase his cable modem.

The issue is the site 'hangs' anywhere from 10-30 seconds, then BOOM! The info flows like the 'Ol Missisip. This behavior isn't hops, network latency, or large pics... something else is afoot.

It also hangs on refreshed of the thread list, forums list, replies, etc..

 

 

Not to contradict dzervit, as he makes a good point, but:

1) The DOS version of tracert shows which hops may have latency problems (with higher transmissios times, timeouts indicated by *'s, etc.), so it is a valuable tool in diagnosing if there are problems between here and there.

2) Requesting a page on this site is not a simple request of one file and one route to the server. There are external files, such as the images Doug mentioned, ads hosted on other servers and Javascript called from other sites. For each and every page request, contact has to be made with several servers, not just ours.

Does our MySQL configuration need some performance tweaking? Yes. But the vast majoriity of connection/performance problems occur between the visitor and the server, i.e. outside hops, and the speed of your provider (ie DSL, cable, dial-up), is almost irrelevant.

To those (many) people that say, "it only happens when I visit this site" ask yourself how many times you visit this site each day compared to other sites...

Incidentally, we have a performance SLA which states that if our latency between our server and the handoff to the internet exceeds 40 ms, we get $'s back. We've never had to exercise that SLA.

FWIW, we've also had people complain about performance and it turned out that they had our old IP address in their hosts file. Some might still have old DNS information cached on their PC's.

 

You know, you make an interesting point. Many items are pulled off of other sources. Banners, java code, images, blah blah blah. But in most cases if your waiting on a single object the page is loaded partially and the objects are filled in as fetched. The delay Jeff & I speak of is almost like a 'pause'. Nothing happens. No attempt at loading data, from anywhere. It's like the server doesn't even exsist. Them BOOM! data comes in. This is not the behavior of network latency issues, and every time it happens I verify I can ping the server. Sometimes yes, sometimes no!

Now, the last thing I want to do is tell you how to run your site, but something just ain't right sometimes...

Anyone up for some simultaneous packet captures?! Crap! What did I do with that software..?!

Oh well... I've lived with it this long.

 

You can ping f150online.com, but did you also try to ping doubleclick.com, google.com, etc, etc. When you request a page from F150online, our web server will open up and maintain a single connection for each and every resource that resides on our server. Resources not on our server are outside of our control. Every page on the site contains resources not on our server....which is exactly my point. There are times when I pull up a page, see just the site header for ten seconds and then the rest of the page loads. The point at which to page is hanging is the Ford ad, served by Doubleclick. Is that ad guilty 100% of the time? No, but it is many times...

 

Ok, looking at the source code of this page, it looks like any other standard web page... my browser pulls the code, and my browser goes out to grap the items from doubleclick, google, etc... the server isn't doing it for me. Yes there is some server-side php stuff, but that isn't the ads or other 3rd party stuff, right?

If that's the case, then my browser loads what it has and fills in the gaps as the data comes in. Instead, I get nothing! So is the server waiting for the code from another site to serve me the code for the page or what? That's what I think your implying, but that wouldn't make sense. Me thinks me needs to study more HTML/PHP...

 

But I wasnt talking about cable modem users - I was talking about the average member ... If everyone had 800 pixel images in thier sig and everyone displyed the sig every time everyones performance would be negativelt impacted. Would the dial up user see the difference MORE than the cable modem user yes ... But that doesnt change the fact that as steve said every little bit helps.

And more to the point a sig pic that big is unneccessary and excessive in my humble opinion. As is posting it in EVERY post - Often 2 or 3 or more times in the same thread...

dzervit - you come to these conclusions based on your years of hosting popular, active mysql, coldfusion based web pages ??

Doug

 

[code]Ok, looking at the source code of this page, it looks like any other standard web page... [/code]When you "view source" of a webpage, what you are seeing is the HTML that is ultimately rendered for the client (i.e. the web browser) to layout the page. It is always going to look like standard HTML (or Javascript, XML, etc) once it gets to the browser itself regardless of the server or programming language it originated from...

 

I must have misunderstood your post then. He stated he had a cable modem and you implied a contributing factor was large pics like his sig.

Posting your sig mulitple time in the same thread should result in a single download of the image, and then pulled from cache for the rest of that thread... Thus, putting a sig in each reply really makes no speed impact on the user. It does make it easier to follow your replies and a thread, though!

My assumptions on the performance issues come from years of network performance & design, application profiling, and information security work. Like I said before, there are tons of variables, and only some in-depth analysis would tell the real tale.

 

To take the point a bit further as a followup to my earlier post...

The way the page is laid out, there are several sections. The header, the banner ad, the body and the footer. The page is designed this way intentionally, to render the page more efficiently under ideal scenarios.

The content of the page loads in that order and will not proceed until the prior section has completed its request to the server.

All of the images in the header reside on this server so when the server "hangs" it is usually after the header has been rendered.

The images for the banner ads reside on both this server and external servers, depending on the ad being displayed. Depending on the browser, until the banner request is completed, further (external) server request and page rendering do not take place.

The next section becomes more complicated, as it uses several embedded HTML tables, external calls to Google and, if users have images hosted on outside servers, calls to other servers. Some browsers will not render the entire page layout until all item requests are fulfilled or they time out and give up trying. (check out Firefox to perceive must faster loading times)

The last section, the footer, is just like the header.

With all that said, I do acknowledge again that we need some performance tuning on the db side, but those problems are the exception, not the norm.