How it works in vB: a search only runs once at the thread post time, and later only if the thread title is edited.
So the similar threads block generally lists only the threads posted before the currently viewed thread post time.
By the way, using Sphinx for similar thread search greatly enhances the relevancy of the found similar threads.
You can estimate the speed of the conversion process by issuing SELECT COUNT(*) against the IPB's post table regularly (sorry, I don't know the exact table name). It would be great if you could share the conversion rate stats.
Let me fix it - there is a 20 million post IPB forum.
Oh yes, tell me about it.
So as I suspected, the difference in big board numbers is because of the underlying forum software popularity.
One must have very conservative approach when running a big board as a business. You can't just switch the forum software at a whim according to the latest trend.
No, it means that vB is good enough to run a big board. It's irrelevant that IPB may be even better - until a sufficient number of forum operators switches to IPB proving that it can handle the load, the situation will stay the same (see above about the webmasters being conservative).
Tell me, were there any big board switches from vBulletin, now that the vB4 debacle continues to unfold? There is much grumbling in the vB community, but I bet the actual switches to IPB in between the big boards are few, if they are happening at all. That's because the big board operators are probably the most conservative bunch of webmasters out there :)
It's a never-ending circle that would be very hard for IPS to crack. Something really disruptive must happen in the industry, like IPB 3.1 suddenly being really really REALLY great, providing everything vBulletin has and then some, combined with VBS continuing to dawdle in vB4 bugs, or even shooting themselves in the foot another time with vB5.
It was an excellent sales pitch. But no one has compared IPB and vB on the same hardware under a considerable load, right? So you can't really guarantee that our forum will continue to run fine on the same hardware but powered by IPB.
If only you had a success story of some big board that happily migrated from vBulletin...
Of course, the board size isn't all - the user activity is probably even greater factor contributing to the server load.
I'm actually not that interested in switching at this time. If we ever enter into the IPB land, it would be through IP.Converge (when it goes open source as in "I can read it") :)
I know about idolforums, it's #1 at Big Boards. It seems to be running IPB 2.x. Are there other 20+ (or even 15+) million posts IPB forums not listed there?
And I'm sorry, that "they had a head start" theory doesn't explain 1:10 ratio of IPB to vBulletin seen at Big-Boards at the top of the chart. For example, there are just 27 IPB forums with more than 5 million posts, and 228 vBulletin based forums with 5,000,000+ posts. An year of head start can't explain that difference. Looks like it comes down to the platform popularity after all.
Our biggest forum is modest by vBulletin's standard, about 15 million posts, but it would be in the IPB top 5 if we converted today...
The reason I'm so curious about the really big IPB boards is that it would be a certain testament to the platform scalability and performance. The more big boards use IPB, the less potential problems one would encounter going IPB on a big board.
Thanks for those big board examples. Do you know any IPB forums with more than 20 million posts?
Big-boards may not be maintained anymore, and the database list may be outdated, but I think IPB and vBulletin are treated the same there. Don't vBulletin board operators have to submit their sites to the database, too?
No, I'm not a customer yet. And looks like it will stay so for some time... until IP.Converge 2 materializes in the unencoded form. I've had enough of the encoded PHP applications in my time. Security through obscurity fails, and not only because of the false sense of security.
My impression of IP.Converge that it is somewhat second- or even 3rd-rate IPS product, being worked on intermittently in between the releases. It certainly doesn't raise my confidence in the product, but even with that, if it was unencoded, I could always look in the source and see what's going on or even fix a bug myself or modify it to my needs. Obviously it's out of question With the encoded binary black box approach.
Raza, thanks for the link. Here's the relevant quote, for the history: