Questions

[BulkBear]

I am interested in the Yatishare script.

I've noticed there is a Amazon S3 plugin and a Wasabi plugin. Why are they seperate? Wasabi is completely S3 compatible, the only different is within the service URI. Doesn't seem sensible to pay $69 for a plugin where only the URI is different? Or am I missing something?

Also can someone explain how S3 integration works back end? Are files landed into a staging area, or passed directly to S3 in realtime?

 

[adam]

I've noticed there is a Amazon S3 plugin and a Wasabi plugin. Why are they seperate? Wasabi is completely S3 compatible, the only different is within the service URI. Doesn't seem sensible to pay $69 for a plugin where only the URI is different? Or am I missing something?

Normally people only need one or the other, hence why they're available separately. If you want both I'd be happy to provide the 2nd for free.

Also can someone explain how S3 integration works back end? Are files landed into a staging area, or passed directly to S3 in realtime?

They're routed via your main site, so I guess a staging area. There's no way to apply things like speed limited downloads, image previews, reading meta data etc if they're sent directly.

 

[BulkBear]

Thanks Adam,

I appreciate the offer there!

So just I am clear about this.... Files are uploaded using a normal http post, then go into some sort of temporary staging folder. From there is there then some kind of cron job which moves them to S3? When the files are downloaded, does the reverse happen, or does the user access them directly from S3?

On a seperate note, does the application work well when load-balanced across multiple nodes? In other words, would it be possible to have 10 web servers, 1 database server, and one S3 storage endpoint?

 

[adam]

Files are uploaded using a normal http post, then go into some sort of temporary staging folder. From there is there then some kind of cron job which moves them to S3? When the files are downloaded, does the reverse happen, or does the user access them directly from S3?

No they're immediately moved to S3, same on download. So the user uploads them to the site, then when at 100% they're moved to S3. The user would be told the upload is complete when it's fully stored in S3.

On a seperate note, does the application work well when load-balanced across multiple nodes? In other words, would it be possible to have 10 web servers, 1 database server, and one S3 storage endpoint?

I don't see why not. You'd need to have shared storage for logs and cache, apart from that it should just work. Sessions can be DB based so there's no issues there.