Visvanath Ratnaweera

Update: The FOSS fans can relax. it's just another partnership, insist GitHub CEO: https://www.theregister.co.uk/2019/05s/24/github_is_independent_despite_microsoft_acquisition_insists_ceo/.

The strategist behind all this: https://www.theregister.co.uk/2019/02/05/satya_nadella_five_years_microsoft_ceo/.

I understand. Note that the lowest VPS and (shared) mass hosting have some overlap. Most of the mass hosting providers offer a whole lot of web applications on them. I know, no command line access! Restricted and stiff for the more advanced Moodle administrator. But might suffice in your case.

Hi James

Sorry, I didn't follow the whole discussion. Mainly because I don't think sharing a VPS to run multiple Moodle instances by multiple system administrators makes sense. Firstly, for a VPS costing USD 5 - 40 per month the admins will lose much more coordinating the group. Secondly, a VPS is already a shared resource, the virtualization doing the partitioning.

On the other hand, informal sharing of a dedicated server makes sense. They are more expensive. But each Moodles instance has access to much more resources, as long as the peak activity of the Moodle instances do not overlap. On top of that it is environment friendly, less energy burnt per Moodle instance.

Hi

You've got expert advice, I can confirm. ;) Imagine, this is Open Source Software and a community forum!

My take on this: The distribution is not the problem. The merge is! There is no freely available solution to merging a collection of global sites back to the on premise site. Apparently Moodle Mobile App has implemented parts of it. But needs to be developed further for the full functionality. BTW, the merge needs to be followed by a new round of distribution, right? Otherwise global sites will drift away from the in premise site.

There are a few things I don't understand about your approach though. Firstly, what is the problem this model is supposed to solve. You said latency to Australia is a big issue. How does it manifest? Have you observed that the UK site is too slow in Australia? How did you measure? Do you have any figures? See https://docs.moodle.org/en/Performance_recommendations for the methods which are know to us.

Secondly, once you have those numbers, how do you know that the distributed model is the solution? It is a radical one. Is it possible that you exchange one problem with many other problems. The Performance recommendations has many suggestions. But for that, you need to find out the cause of your problem.

Footnote: The above are the same "slightly unhelpful comments" here https://moodle.org/mod/forum/discuss.php?d=386620#p1558472.

Thirdly, you seem to know the technology needed, Microsot Azure, before doing the analysis(?
)

Well, since you've engaged a professional, no need to answer all those questions yourself. It is just a watch list AFAIAC. Of course it would be very nice if the result will be published under a Free License. There is more than one taker: https://discuss.moodlebox.net/d/27-a-pool-of-moodleboxes-an-idea ;)

Hi

If you are familiar with Linux and considered that option, then I said nothing! ;)

But then _caution_: I didn't suggest you to migrate tomorrow. Such a major change needs planning. The earliest I can think of it the summer break (if you are in the northern hemisphere).
Saw that Ken gave a longer answer.

The main point, I don't say Windows make a bad server for Moodle. But you see the problem. You have two Moodle servers doing fine on Windows and one not. But your debugging capabilities are limited. Or, to put it another way, your moves are dictated by Windows, not the vast choice Linux offers. (Sorry, I can not answer your particular question, I migrated to Linux decades ago! Even my desktop is Linux.)