"the state of the linode"
i'd love to see an annual commentary about the state of linode. past, present, and future stuff.
what's prompting me to write in this moment is my curiosity about linode's future wrt sun containers. i'm wondering if you think it will be difficult for uml or xen to compete against sun's containers + zfs in the area of features/dollars?
one isp that is embracing sun's technology (software and hardware) is textdrive.com (a.k.a. joyent.com):
they are planning on offering smaller containers in the future.
joyent will be using sun's technology for more than just vps services. they will be using it to power their storage solutions (strongspace and bingo), their application services (joyent connector), and even their shared hosting solutions. it will be powering pretty much their whole backend.
here are some glimpses into the details:
so, do you see sun's technology as a potent competitor? and if so how do you plan on competing against it?
The Sun solution has got to be more expensive than commodity server equipment, look at textdrive's pricing. They're nearly twice what's going around in our market. I'm sure that'll come down, but so will we/everyone esle.
Also, Xen + shared storage can do all the neat things Containers+ZFS can – live migrations, etc.
As far as competition, we're going to keep doing what we've always done.
In Solaris 10 there are fatal limitations on how memory and I/O resources can be restricted inside each zone. It can handle CPU fairly.
I have no idea how much more advanced opensolaris is over solaris 10, but it'd have a lot of handicaps to overcome.
For true virtualisation Sun is looking at Zen on intel/amd platforms and LPAN on the Niagra Sparc CPUs.
OpenSolaris (Solaris Express) may have the new rcapd already; the official Solaris 10 release trails OpenSolaris by a fair few months.