The CDN may not be ideal for everyone. There can be a bad route to any of those locations as easily as to any server. If you're seeing majorly decreased speeds I encourage you to try the other test locations... I know now why you and others might not be able to see the central.testmy.net server... I had the 3rd and 4th nameservers set to my old ns. I corrected this, you should be able to see that soon.
By the way... I've got a really major addition I'm working on. I've got a brand new server up in what is considered by most to be the most powerful datacenter that Softlayer has. Dual Quad-Core (Xeon-SandyBridge E5-2620-HexCore) 16 GB RAM (split across 8 RAM slots)... 4 200GB SSD in RAID5 array... 2000 Mbps uplink. The current server is powerful but this new one is exponentially faster. RAID5 with those SSDs is VERY fast and reliable. Server is up and burnt in. I've just got a lot to setup and transfer... but I should have it ready for production VERY soon.
Back to the CDN. I'm using what's called "Origin Pull". The CDN routes TestMy.net's DNS to the closest server to the user. It then relays the request. TestMy.net sends the request to the CDN and the CDN sends it to the user. Time is taken on the client-side so the fact that the request is being relayed doesn't effect the calculation.
... Some will see improved speeds with this method some won't, it's just another way to test. I may put the default server back to Texas, I still think that a longer route is a better benchmark. But the fact is, more and more content is being delivered this way. I was really impressed with the way that Cloudburst worked with my application. But I'm also working with Google's CDN and a few others to provide more testing options. What better way to test Internet performance than through the networks that are actually delivering the content...
I just can't wait to get the new home server finished. It's a big process, I have to make very large database transfers quickly and get them switched on to the new server before I can even start setting the actual domain up. Otherwise more posts, signups and scores will be lost. I say more because there's going to be at least a little gap in the database transfer process. Most people won't notice, but there may be a 5-25 minute window of time that will get lost. But once the bandaid is ripped off it will be nice.
Obviously once this is online I'll post.