I think it's way less than 2 years. Reminds of the time they (as online.net) hiked the prices of their lost-leader dedi boxes by a similar margin back in 2018, they did that in 2 steps just months apart too.
Hah, I saw something about decreasing hourly rates (sounds good) but didn't figure out that they are increasing monthly rates. I should have guessed. If they do spot pricing of hourly bare metal, that will be quite interesting. I tried a Hetzner 16 vcore shared-cpu instance and it was disappointingly slow. Their dedicated resource instances are presumably faster but cost an awful lot more.
Any idea what Scaleway's policies are for sustained CPU usage? I'm not posting hyperscale clouds as they are obviously expensive, but other dedicated core offers aren't the cheapest either, example: https://www.linode.com/pricing/
I am not a Scaleway customer so I believe cannot write them an email (what is wrong with those French companies, OVH also wants me to call lol), and I couldn't find a clear explanation on their website - https://www.scaleway.com/en/pricing/ - but assuming your CPU usage is not throttled, it might still be not that bad (saying this based on what people posted on geekbench.com)
I benchmarked their smallest DEV instance a while back (see the YABS thread) that was 'okay' for 3 euros a month, but definitely not for 5 euros a month and hell no for 7.50 euros a month.
Sorry if I wasn't clear enough, I was trying to make the point that in use cases featuring rather constant high CPU usage, then (and only then) it might be a good idea to use Scaleway - provided they'd allow it (i.e. that these indeed are dedicated CPU or at least close-to-dedicated CPU instances).
(Also, I guess your YABS showed a more complete picture - i.e. it featured disk and network performance too? - whereas AFAIR Geekbench is only about raw computing power)
The links above do prove your point, that is, at first glance, Scaleway's VPS performance look worse when compared to similarly priced VPS at some major providers like Hetzner, Vultr, DO or AWS LIghtsail (presumably all of which managed to provide better results during a temporary burst).
But then again, all of these won't let you have sustained high CPU usage (well, technically, Hetzner and AWS will, but only on plans costing much than Scaleway DEV/GP virtual instances). Instead, they'll either tell you to stop / kick you out (because of the fair share clause inside their TOS - Hetzner, and most likely DO [their TOS are somewhat ambiguous IMO]) or just throttle you (Vultr/AWS Lightsail - in case of the latter, your guaranteed CPU share is somewhere between 5 and 40 percent per core, depending on the plan).
If you're using shared cpu instances then even with no throttling, performance will fluctuate because of other users also computing. Some people's EC2 orchestration involves automatically running a benchmark when they spin up a new server, and if it is slow, spin it down and spin up a different one until finding one that is unloaded.
Even the "dedicated vcore" instances will vary, since on an unloaded machine a vcore is basically a full pcore, while on a loaded one it's about 60% of that.
I do like Scaleway's idea of having hourly pricing of bare metal servers. Hetzner currently has nothing like that. OTOH their auction servers are so cheap now that you're often better off just keeping one of those spinning.
@AnthonySmith said:
You have to wonder what customer retention is like when companies do this.
The thing is it is usually only 10% of the people that make the big noise, so if only 20% leave, they win.
I think it's normal for every corporation to try and boost their profits - which includes price hikes as well. Why invest corporate money and effort in building a brand if it doesn't bring more profits?
Pay attention to such people in life:
They don't increase prices, they harmonize prices.
They don't fire people, they consolidate the company.
They don't pay low, they offer competitive salaries.
They don't lie, they offer objective truth.
They don't kill, they neutralize threats.
They don't sway, they guide.
etc.
There are honest people here on the market, also working for profit, that are not doubletalking manipulative sociopaths.
Comments
I think it's way less than 2 years. Reminds of the time they (as online.net) hiked the prices of their lost-leader dedi boxes by a similar margin back in 2018, they did that in 2 steps just months apart too.
I benchmarked their smallest DEV instance a while back (see the YABS thread) that was 'okay' for 3 euros a month, but definitely not for 5 euros a month and hell no for 7.50 euros a month.
LinuxFreek.com
Sorry if I wasn't clear enough, I was trying to make the point that in use cases featuring rather constant high CPU usage, then (and only then) it might be a good idea to use Scaleway - provided they'd allow it (i.e. that these indeed are dedicated CPU or at least close-to-dedicated CPU instances).
(Also, I guess your YABS showed a more complete picture - i.e. it featured disk and network performance too? - whereas AFAIR Geekbench is only about raw computing power)
The links above do prove your point, that is, at first glance, Scaleway's VPS performance look worse when compared to similarly priced VPS at some major providers like Hetzner, Vultr, DO or AWS LIghtsail (presumably all of which managed to provide better results during a temporary burst).
But then again, all of these won't let you have sustained high CPU usage (well, technically, Hetzner and AWS will, but only on plans costing much than Scaleway DEV/GP virtual instances). Instead, they'll either tell you to stop / kick you out (because of the fair share clause inside their TOS - Hetzner, and most likely DO [their TOS are somewhat ambiguous IMO]) or just throttle you (Vultr/AWS Lightsail - in case of the latter, your guaranteed CPU share is somewhere between 5 and 40 percent per core, depending on the plan).
Contribute your idling VPS/dedi (link), Android (link) or iOS (link) devices to medical research
Looks like it is time to move on.
Farewell Scaleway!
If you're using shared cpu instances then even with no throttling, performance will fluctuate because of other users also computing. Some people's EC2 orchestration involves automatically running a benchmark when they spin up a new server, and if it is slow, spin it down and spin up a different one until finding one that is unloaded.
Even the "dedicated vcore" instances will vary, since on an unloaded machine a vcore is basically a full pcore, while on a loaded one it's about 60% of that.
I do like Scaleway's idea of having hourly pricing of bare metal servers. Hetzner currently has nothing like that. OTOH their auction servers are so cheap now that you're often better off just keeping one of those spinning.
Pay attention to such people in life:
They don't increase prices, they harmonize prices.
They don't fire people, they consolidate the company.
They don't pay low, they offer competitive salaries.
They don't lie, they offer objective truth.
They don't kill, they neutralize threats.
They don't sway, they guide.
etc.
There are honest people here on the market, also working for profit, that are not doubletalking manipulative sociopaths.