OVHcloud's definition of modern is elastic. I've seen Sandy Bridge in their Kimsufi line labeled as "reliable" which is one way to describe a twelve year old Xeon.
That said, for a web host it genuinely does not matter. Your bottleneck is PHP-FPM workers or database locks, not AVX-512 throughput.
Thank you @GeorgeNmp and @josemcelroy for the explanation. I feel less stupid now 😅
I check Vultr price for Paris datacenter, it is 6$ for 1GB RAM. Contabo is 4.99€ for 8GB but with setup fee. For my small website I think I stay where I am.
The E5 v3 is maybe old but my ping to Lyon is 12ms from Gravelines. Hard to beat that.
12ms? I get 180ms to my Contabo Singapore node from Manila on a good day. You Europeans don't know how good you have it.
Member
My 1% utilization would not notice if they ran it on a Pentium 4. I am the ideal customer for EOL hardware.
Member
Everything is about positioning. Your CPU generation is invisible. Your domain is your brand.
But fine, ignore the free alpha. Back to your scheduled obsolescence panic.
Thread is about whether EOL CPUs matter for VPS users. Not domains, not positioning. Please stop derailing.
Following this. I am on Contabo NYC and curious if they will migrate us to newer gen or just run the nodes into the ground.
They will run them until the failure rate exceeds the cost of replacement labor. These are depreciated assets. The sweet spot for a cloud provider is when the hardware is fully written off but still drawing paying customers.
Contabo's model depends on density, not speed. E5 v4 with enough RAM and NVMe cache layers can still saturate 1Gbps. That's their target market.
Same here on Contabo Seattle. I run lscpu and get E5-2630 v4. For my use case it is fine.
