제 데이터베이스는 23 GB이며 포럼은 매일 활동적이지만 특정 분야(niche)에 특화되어 있습니다.
제 성능 테스트는 매우… 거칠다고 할까요, 하지만 새로운 플랜이 제 포럼에 적합한지 확인하는 데는 충분할 것 같습니다
홈페이지를 20번 새로고침하여 중앙값 로딩 시간을 계산했습니다.
CPX31
CX43
중앙값 로딩 시간
523 ms
876 ms
실제로 포럼을 탐색할 때 더 느리게 느껴집니다.
CX43의 더 높은 사양(RAM과 vCPU가 두 배)에도 불구하고, 낮은 가용성 이 로딩 시간에 상당한 악영향을 미치는 것 같습니다(사용자로서 가장 중요한 부분입니다). 시간이 지남에 따라 변동될 수 있으므로 당장 다른 플랜으로 변경하지는 않겠지만, 그 동안 이런 사실을 알아두는 것이 항상 좋은 것 같습니다.
Interesting. My home page load is about 400ms. I’m using CAX11, which is ARM based, and has only 40G storage. My compressed forum backup is only 1G. So, a much smaller forum, therefore perhaps faster load.
I would guess ‘limited availability’ means that Hetzner do not have many machines of this spec available. I wouldn’t expect any implication about performance.
Because these shared cloud plans divvy up CPU resources, they are ideal for development and testing environments, blogs, forums, CMS, small databases, and VPN servers.
Cost-efficient on older hardware generations, with limited availability. Ideal for variable workloads that can handle changing CPU performance. Optimized for low CPU usage with variable workloads.
(backend)
Optimized for cost efficiency, available only in limited capacity.
One thing about shared machines - and they always are shared - is that the other tenants might be very quiet, or very active. If they saturate the i/o, or use lots of memory bandwidth, you get a fair share, but that’ll be less than if they were quiet.
Yeah let me say something I’m watching a forum that using cloud too so maybe this be the big trouble for high traffic specially them that is was had many outages this month
I’m self-hosting using bare metal + virtualizor, you could try OVH or idk
That’s right. I know how it works, but I often forget it when looking at performances.
Out of curiosity, I measured reloading times to see if it changed, since the last time, but the median loading time is still the same (874ms Vs 876 last time).
I’ll try to move to a CPX32 (same specs as the discontinued CPX31, but cheaper, and as I heard, slightly better perfs) to see how it compares
제한된 가용성
