Hello, I'd like to ask a few questions.
I'm not very technically skilled and would like to apologize beforehand if these questions seem too trivial.

1) If I get more than one VPS in the same datacenter are transfers between the VPS "out of band" ?

2) Since it's possible to choose between three different datacenters I'd like to know if there are notable differences among them besides the geographical location.
For example I read in another post something about different behaviour regarding "filtered ports", I don't know if that was meant as closed ports or some kind of connection throttling.

If possible I'd like to know an IP address for each datacenter so I can test latency.

3) Do "bigger" plans get more IO tokens? Are they directly proportional to the allocated disk space?

4) Is it still possible to ask for migration on a xen box? Is this tied to a particular datacenter?

5) Is it possible to recompile the kernel with UML?

6) Is the RAM you advertise in your plans really available physically?
Sorry if this one sounds quite dumb, I don't even know if RAM overselling is possible with UML.

Thank you!
-- jvm

Hello,

Thanks for your interest in Linode.

1) Currently all bandwidth is accounted for, but we're working on support for private back-end networks between Linodes within the same datacenter that will remove this limitation.

2) In my opinion, I think all three data centers are great. Atlanta does block some ports, but the only one we ever get complaints about is IRC (port 6667). Most IRC networks have alternate ports you can connect to (8000 for freenode, 6668 for oftc).

IPs to test latency, and files to download:
http://www.linode.com/forums/viewtopic.php?t=2636

3) Larger plans do not get more IO tokens -- the reason for this is that the token IO limiter is only there as an "emergency brake" for runaway Linodes. You should not hit the limiter under normal operation.

4) We've got Xen beta boxes in all three datacenters

5) You can compile anything on your Linode, including kernels, but you can only use the kernels we provide. The kernels we provide have all of the useful stuff enabled (tuntap, a lot of the iptable stuff, etc). Under Xen, you can compile your own modules.

6) Since UML is a (set of) processes on the host, the host can indeed swap them out, however, we never oversell. The full amount of RAM allocated to your Linode *is* physical RAM and will never be swapped out.

Hope that helps,
-Chris

Chris,

thank you very much for your prompt answers!

Could you please confirm that I can ask to have my linode(s) migrated to a xen box anytime? Do you plan on closing the xen beta in the future?

If I understand correctly similar plans are hosted on the same machine.
If you can disclose this, I'd like to know the specs of the machines that host the Linode 720 plans on xen.
If you could also tell me the max VPS/machine ratio for the above plan that would really be awesome!

Thanks a lot!
--jvm

You can ask to be migrated to the Xen box at any time, however eventually they will be filled. My hope is that by that time Xen has proven itself (it's been a long road with Xen, but I'm feeling good about it finally) and we'll start to bring up new machines under Xen.

We group similar plans onto the same machine. The guest:host ratios for the plans are: 40:1, 30:1, 20:1, 10:1 for L360, L540, L720, L1080 respectively.

The Xen hosts are SuperMicro 1Us with Quad Core Xeons (x2), 3ware hardware RAID1 with SATA2 hotswap drives, LotsaRam<tm>.

Thanks!
-Chris

Well, then I'd say it's time for me to choose a datacenter

Regards,
jvm