If you're lucky and your machine supports coreboot there is even better way to boot into linux - use slimmed down linux kernel as a direct payload of coreboot and kexec into your actual kernel. This way you're skipping any disadvantages of grub (if you can name any which matter) and don't need to deal with relatively new and not as much battle tested code to boot your trusted kernel.
No, just an intermediate kernel is written into flash memory along with coreboot. It doesn't wear little flash chip as it doesn't write anything which needs to persist, that's actually one of the points of this solution. Its sole role is to chain-load your destination kernel. So you just basically move your bootloader into non trivially writable memory (as it can be write-locked) to make it more tamper proof.
There is always a risk of bricking your system if you're not careful, but wearing off flash isn't one of those risks if that's the concern.
Nope; bootloader flash chips use NOR instead of NAND flash precisely to prevent this.
NOR flash guaranteed for a minimum of 100,000 complete rewrites of every cell, with no flash translation layer gimmickery. The downside is that it is very expensive, which is why a few dozen megabytes of NOR flash costs as much as many gigabytes of MLC NAND.
Typical NOR flash chip; "Minimum 100,000 Program/Erase Cycles":
NAND flash vendors often advertise "up to" a similar number of P/E cycles on their SLC devices, but you'll have a hard time getting them to guarantee that. And for TLC and MLC NAND you find in non-industrial products the number is more like 3% of that:
RIP Linode. It was the 2nd (after Slicehost) "cloud service" I was introduced to back then in 2009 I believe, when my, back then CTO, insisted that "cloud" is the way to go and some job skills (sysadmin) will eventually fade and turn irrelevant. 3 years later same CTO complained that I made other teams within that startup look bad by using Hetzner and OVH duo instead of Amazon's then still relatively new AWS - bills were telling the story way better than I could!
Basically the same here. I used Slicehost until they were acquired by Rackspace... and for a while afterwards, truth be told. Inertia is a real thing, let me tell you.
After Slicehost/Rackspace, my next VPS provider was Linode. To be fair, I do also use Amazon for some stuff, but mostly just transient instances, and limited use of some of their more specialized services like SNS.
I experimented with Hetzner and OVH, and wound up deciding to migrate my stuff to OVH. So far so good, but as mentioned in another thread, while I'm fairly happy with OVH, there are a few little "gotcha's" here and there.
