I have two ubuntu trusty instances in dream compute running our multisite network - one is a private ip dedicated for the mysql database, the other is public running nginx and serving the wordpress install. I'm running the W3Total Cache plugin with Varnish, memcached, and APC. I set this up and made several configuration changes following many tutorials and tests and learning a lot along the way.
We've also just started using dream objects with dream speed to store some subsites static assets off our servers. We are using the dreamspeed plugin created by ipstenu.
We have over 40 sites and several of them are asset heavy. They are going to balloon up quickly with several years of archives of images and podcasts being uploaded to those sites.
I was hopeful that dream objects would relieve our primary server from having to store all of those assets, but it seems the plugin doesn't exactly work that way. We can copy assets to the bucket and serve them from the CDN but those files remain on our server and take up disk space. If every site upload is going onto our main server the volume needs will continue to snowball.
I am wondering what is the best way to scale this setup? I've considered:
-some way to symlink the /wp-content/ folder to an elastic volume like dream objects?
-re-size the volume attatched to our main server instance. (Temporary fix)
Has anyone figured out how to make this work?