3 ms·

The 8$/month machine sounds very interesting. I'm quite new to VPS. I recently deployed a couple of project for customer in Linode. I am very happy with them,

The 8$/month machine sounds very interesting. I'm quite new to VPS. I recently deployed a couple of project for customer in Linode. I am very happy with them, but at the same time, I have some hobby projects of my own that I dont want to host in a 20USD/month machine. How are your VPSs compare to Linode?

lsc 16y ago
they have better CPU than I do (and this is unlikely to change... but you know? very few things my customers do are CPU bound.) I also give you fewer VCPU on the lower-end instances. they give you 4, I give you 1. the idea here is to optimize for the worst case (if there is contention, and you have 4 vcpus, that's more context switching and more cache flushes. on the other hand, cpu is usually mostly idle, and in that case more vcpu is more performance. My large instances have more vcpu.) I'm experimenting with my vcpu policy and may bring that more in line with linode. But really, I regard this as relatively unimportant because VPSs are rarely CPU bound. Our disk quality is likely about the same (I use 7500rpm "enterprise" sata, which is about as good as sata gets without getting into the expensive 10K stuff, and if linode was using 10K disk, they'd be advertising the fact.) My ratio of disk to ram is 4 disks in raid 1+0 for every 32GiB ram/8 cores I'm not sure what Linode's ratio is. But Linode has better mechanisms for limiting heavy disk users than I do. (heavy in terms of performance rather than in terms of space... limiting by space is trivial.) My understanding is that they keep iop counters or similar, and first warn, then limit you if you use a lot more iops than most people. I don't do this, and sometimes overall disk performance suffers because of it. But I do plan to emulate Linode in that regard at some point. And, of course, my most noticeable deficiency is that I do manual provisioning, and Linode has a nicely automated setup. but I'm (very slowly) catching up. Now, I'm about half the price, too, but you already knew that.

lsc 16y ago

they have better CPU than I do (and this is unlikely to change... but you know? very few things my customers do are CPU bound.) I also give you fewer VCPU on the lower-end instances. they give you 4, I give you 1. the idea here is to optimize for the worst case (if there is contention, and you have 4 vcpus, that's more context switching and more cache flushes. on the other hand, cpu is usually mostly idle, and in that case more vcpu is more performance. My large instances have more vcpu.) I'm experimenting with my vcpu policy and may bring that more in line with linode. But really, I regard this as relatively unimportant because VPSs are rarely CPU bound. Our disk quality is likely about the same (I use 7500rpm "enterprise" sata, which is about as good as sata gets without getting into the expensive 10K stuff, and if linode was using 10K disk, they'd be advertising the fact.) My ratio of disk to ram is 4 disks in raid 1+0 for every 32GiB ram/8 cores I'm not sure what Linode's ratio is. But Linode has better mechanisms for limiting heavy disk users than I do. (heavy in terms of performance rather than in terms of space... limiting by space is trivial.) My understanding is that they keep iop counters or similar, and first warn, then limit you if you use a lot more iops than most people. I don't do this, and sometimes overall disk performance suffers because of it. But I do plan to emulate Linode in that regard at some point. And, of course, my most noticeable deficiency is that I do manual provisioning, and Linode has a nicely automated setup. but I'm (very slowly) catching up. Now, I'm about half the price, too, but you already knew that.