Were still using HTTP from an AFP share on our Mac servers.
We need failover, load balancing & scheduling of replication.. Which is why we've not made the move.
Why exactly are my thoughts, I think it may appeal to some but personally I'm with Ben on that. When dealing globally failover, load balancing and scheduled replication is a must
We're running vanilla JDS', but are working on setting them up with AnyCast IP Addresses, so if one goes down the client will redirect to a nearby JDS. We also have an Amazon CloudFront-based CDS, so if I know a region is going to loose it's JDS I can redirect clients to the CDS for a time. Not perfect, but it works.
@kilodelta what a great idea :)
For me, the big plus for JDS is its replication and throttling capabilities as well as its stupid simple setup on OS X and Linux.
In my old environment, I had 12 global sites with WAN links back to our central office. My network folks jumped all over me when I'd copy something like an Adobe Creative Suite package across the WAN to my AFP/SMB distribution points. There's goes the bandwidth!
While it would have certainly taken a while, I could have used the throttling capabilities of JDS to distribute my packages across our WAN in a friendlier manner.
A JDS distirbution point is not for everyone. I still recommend a regular AFP or SMB file share for most customers I support. But it certainly has its place in a dispersed network.