I think the board is slowing down because of the way pages are scrolled out now.

They used to pop up all at once.
Now they come in a lot of chunks.
This is supposed to make it easier to start reading while the transfer is still going on.

The effect, though is more separate transmissions. That seems like it would have to be slower overall.

I don’t much about this, but no one else has noted that this is the major change of the new modifications.

Just thought I’d see if it rang any bells with people who do know about this.

I hear bells ringing, but I don’t know if it’s tinnitus or the Taos hum.

The pages scroll differently because of the wild fluctuations in thruput experienced by the server.

It’s a symptom, not a cause.

However, Jerry the Tech Guy (our hero) has made things better; the board is back to having the kind of thruput it did before we applied the upgrade.

So let’s see how this works . . . still a work in progress, every bit of it . . . but I’m cautiously optimistic.

your humble TubaDiva
Administrator

Actually Tuba, the pages are loading differently now. Before, the program creating a page would send the header info, then put the whole of the thread together before sending that (then the rest of the page, but that usually didn’t take long) Now, the program sends each post separately. If you don’t believe me, find a large file to download, then view a long thread. The thread will be loaded post by post, rather than the waiting for the whole thread to be put together.

I like it - if there is a thread with a bunch of posts already that I haven’t read, I can be reading the initial posts while the rest of the thread is loading. It can however have a negative impact on the network - because of the way TCP/IP works, it can mean more characters are being transmitted. Obviously this is not the primary problem we were having, because this feature hasn’t changed and the board is screaming now. However, if Jerry gets bored, he may want to check to see if he can tweak the system to load the posts in 2 - 5 post increments, instead of one at a time.

Thanks for the information.

I’ll pass this on to Jerry as well; doubtless he’s aware of what’s going on, but never hurts to make a mention.

I apologize for anything I said in ignorance.

your humble TubaDiva
Administrator

Version 1.1.0 of vBulletin placed the entire post section of a thread in one large table. In version 1.1.3 each post is contained in it’s own table which means that the threads have indeed been “chunkized” per post and will render that way. This really doesn’t put any more load on the server. HTML is just ascii text streaming over the internet.

The server isn’t really sending any more data packets unless the overall size of the HTML has increased (which it probably has…but only minimally). The real effect is that your local browser renders the page in a different way. Basically your browser no longer waits for the entire large table with all the posts in it to be received before rendering the posts but renders each individual post as it arrives insteads.

It’s pretty much a win/win situation since the server isn’t really working any harder and the browser is showing partial contents of the thread faster.

Jerry

I think I missed “chunkized” in my MCSE exams. That’s one of them specialized techie terms, ain’t it? :smiley:

Thanks, Jerry. Did I mention I have a shrine to you in my office?