Suggestion for reducing bandwidth woes

[Arthmoor]

Something I saw over at theadminzone.com that caught my eye - signatures get displayed once per user per thread. Someone asked why they did it and the answer was that it cuts down on used bandwidth. Now, they don't seem to even allow sigpics, but the logic is sound. Perhaps something like that would be useful here? I'm not sure if IPB has a config setting for that but it seems that if each sig is only displayed once in a thread instead of many times that you'd save quite a bit on bandwidth, and if done right perhaps even in data queries.

[Dark0ne]

Once the image is downloaded once it is stored in the user's browser cache, from which the image is called if it is displayed again on that page, and normally after reloading the page etc.. Ergo only showing the sig image once will have no bearing on bandwidth to the user or the site, as the image is only downloaded once.

[Arthmoor]

Well even if the cache on the user's end is functioning properly, it would still do a good job of cutting down queries as well since the DB won't need to grab the sig for every single post on a thread that way. I don't know exactly how it works but TAZ seemed to think it was well worth it.

[Kodiack]

Well even if the cache on the user's end is functioning properly, it would still do a good job of cutting down queries as well since the DB won't need to grab the sig for every single post on a thread that way. I don't know exactly how it works but TAZ seemed to think it was well worth it.

 

This wouldn't help much. At all. Besides, most images aren't even hosted on this site or server. Also, as Dark0ne already stated, images are cached into users' memory and hard drive. This would lighten up a few bytes of data at most, which even a dial-up connection can handle in less than a fraction of a second.

 

Bandwidth also isn't the only issue. There's also problems with demanding tasks, file access speed, etc. Some tweaks could cut off a few milliseconds of time, but for the true performance benefits big things - like using the automatic SQL table optimizer - would provide the largest improvement.

[Arthmoor]

A few milliseconds of time magnified by 750,000 users can add up to a lot of freed up resources.

 

The optimizer would surely help, but it isn't a magic bullet either when you're dealing with so many people.

[Kodiack]

A few milliseconds of time magnified by 750,000 users can add up to a lot of freed up resources.

 

The optimizer would surely help, but it isn't a magic bullet either when you're dealing with so many people.

 

Actually, very few resources would be freed up. So few, in fact, that it's really not feasible. :P If you've looked at the back end of IPB - or any forum software for that matter - you'd notice that there's already a lot of stuff being handled. It's amazing how much a server truly does, and removing a few bytes of data from each page load would do virtually nothing to performance.

 

I'm guessing that Dark0ne and/or buddah already work with keeping the SQL tables optimized, but if they don't, it would help a lot. I help run a rather large site myself that also had some performance issues, and believe me, having everything oiled there helps quite a bit with making the queries work efficiently.

[buddah]

A few milliseconds of time magnified by 750,000 users can add up to a lot of freed up resources.

 

The optimizer would surely help, but it isn't a magic bullet either when you're dealing with so many people.

 

Actually, very few resources would be freed up. So few, in fact, that it's really not feasible. :P If you've looked at the back end of IPB - or any forum software for that matter - you'd notice that there's already a lot of stuff being handled. It's amazing how much a server truly does, and removing a few bytes of data from each page load would do virtually nothing to performance.

 

I'm guessing that Dark0ne and/or buddah already work with keeping the SQL tables optimized, but if they don't, it would help a lot. I help run a rather large site myself that also had some performance issues, and believe me, having everything oiled there helps quite a bit with making the queries work efficiently.

 

Dark0ne handles all that kind of stuff.....too many cooks spoil the broth attitude. I just make suggestions and recommendations for improvements. Because the sites are some what symbiotic in nature...there is some rather unique coding involved. In this area I have no skills, my understanding of how the internet works is my strong point. That and an excellent memory...I spend many hours on line reading and trying to stay informed on what is going on here and at the official forums.

 

I actually do read almost every file description and download every file. Two reasons are to keep the site out of trouble (legal) and an unofficial off line archive of the files...I could restore over 95% of the files of the site if necessary.

 

When TesSource died due to legal reasons and TesNexus arose from the ashes, LHammonds and myself reuplooaded over a thousand files to make them available again. In fact with the help of the community all but about 25 have been restored. Worm80275 just uploaded 7 that had been lost last weekend.

 

He has much he wants to do, but he is still in college (university) and time is a constraint that cannot be changed.

 

He reads these posts several times a day, and has a list of what is needed and does them one at a time to keep the site up as much as possible.

 

Micky

[dezdimona]

A few milliseconds of time magnified by 750,000 users can add up to a lot of freed up resources.

 

The optimizer would surely help, but it isn't a magic bullet either when you're dealing with so many people.

I somehow doubt that all registered users are on-line at the samr time given the location of the site and locations of the registered users worldwide.

 

I think,tell me if I', wrong that removing pages worth of dead topics may cause an improvement?

[Kodiack]

I think,tell me if I', wrong that removing pages worth of dead topics may cause an improvement?

 

It wouldn't. Those take up space on the site itself, but they're rarely accessed, and when they are, it's the same bandwidth. Unless we are low on storage space, we'd really have no reason to auto-prune stuff. There are a few things that normally can cause sites to be slow:

 

•Excessive CPU usage (not exactly likely here, but still a possibility)

•Low bandwidth (possible, but not the primary thing to look at here)

•Slow file access (very possible; this is what the optimization would help resolve)

•Extremely high memory usage (unless we're running on some ghetto server, I doubt it)

•Large, bulky pages that are slow on the users' end (definitely not the issue)

 

So yeah, Dark0ne would want to keep an eye on server resources, server load, etc. This can all be done easily via the ACP, although it cannot check every little detail. Its primary uses are for checking CPU usage, memory usage and the server load, so it's not exactly perfect.

[Vagrant0]

I think,tell me if I', wrong that removing pages worth of dead topics may cause an improvement?

Might not free up bandwidth, but might free up some storage space.

 

But the biggest problem with that sort of thing is that it would effectively remove parts of the knowledge base in several areas of the site. There is nothing more annoying than having to track down some obscure bug or script usage, only to find that that thread no longer exists over at Beth forums. While yes, some areas could probably benefit from some pruning, it would have to be done in an intelligent manner so as to not remove anything of moderate importance. Meaning that it would have to be done manually, meaning that one of the poor moderators would have to spend quite a bit of time going through thread by thread trying to determine useful from not... All for the sake of a few hundred mb. Frankly, you could achieve more just by removing a handful of image uploads which are deemed inappropiate or senseless, or instituting some file size limits on image uploads.