ABR Forum - Pay As You Go?

[Rhys]

There's a bottleneck somewhere. If it hasn't been looked into and identified already then it needs to be, then if investment is needed in new infrastructure to fix it, then tell us what it needs and see how many people would be prepared to chuck a tenner in the pot to help fund it.

I'd make a donation to keep the site running, but If you made it subscription based it will just die a death, there are so many other free bike forums.

That said I've just done a DNS lookup on the domain and it looks like the site is hosted out of 1&1's datacentre in Baden, Germany. They don't tend to limit bandwidth and if there were a comms bottleneck it would be down to 1&1 to fix it. I suspect therefore that Alun is paying for a hosting service on an Opex basis and the site has reached the limits of CPU or RAM that he us paying for and needs to pay for more resource. It's difficult therefore from a cash flow perspective to fund an increase in Operational costs using a donation system which essentially gives you sporadic Capex budget.

I still don't think the site should be subscription based, but I can see why not having a subscription based income can be a barrier to increasing the spend on this site.

The site and forum runs on Joomla which is a open source CMS. It's not designed for this purpose on this kind of scale. Joomla is crap at the best of times even running at a much smaller scale unless the people looking after it really know what they are doing to get it to perform optimally.

Simply getting more resources isn't going to help. It's a poor choice of software and it's poorly implemented.

[Bernard Smith]

So maybe don't fix what ain't broke?

+1

I've taken my laptop back and had windows 7 put on because 8 is too brilliant for its own good and it sucks.

A super +1 about that - Microsoft/Foot/shot...... me too re. 7 tomorrow.

[-Ralph-]

So what does Joomla typically suffer from? I cant imagine this site runs massive SQL queries, if the style sheets were too big I'd expect it to run slow all the time (and only on limited bandwidth). Http requests could be an issue, but again wouldn't that be a permanent issue?

Its the sporadic nature of it that led me to assume a resource or infrastructure issue. I haven't noticed any correlation between poor performance and the number of concurrent users (but I only have the bottom if the page to see the number inline). Perhaps it has multiple virtual cpus and cant cope with being virtualised? ie: every time the host reschedules resource it context switches. Maybe if it has multiple cpus reducing the number would actually improve performance?

[Redmurty]

my recent record is 194 seconds for a pages to load :pinch:

cheers Spud