I am looking to get another server. The disk performance is what is limiting me from using larger servers, since they run out of juice a lot faster then memory and disk on my RS2000. Since CPU and memory isn't my biggest concern I'm thinking maybe try VPS instead of RS. I'm thinking of RS1000 or VPS1000. I have tried looking through the benchmark page but it's not clear to me what kind of difference in perofrmance to expect? What is your experience?
VPS or RS
- Zulan
- Thread is Unresolved
-
-
What kind of performance are you looking for? Throughput or IOPS?
-
As for disk performance, below the test results of my VPS 1000 G11 during normal operation. Now compare to your RS2000. I don't have a RS G11 right now. I guess the most important difference should be computing power. My test results with an older VPS were years ago was much worse than that, an older RS G9.5 was much better than that at 4K blocksiue, but only slightly better at 64K, 512K and 1M. Then again, most of the files that are used a lot are cached in RAM anyway in my case. Obviously, this depends heavily on the application the server is planned to be used for. My credo is: There's no substitute for RAM - besides more RAM! Well, for now I'm stuck with 8GB, but it's enough for my application.
Code
Display More# curl -sL yabs.sh | bash -s -- -ign # ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## # # Yet-Another-Bench-Script # # v2024-06-09 # # https://github.com/masonr/yet-another-bench-script # # ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## # Fr 13. Sep 17:20:52 CEST 2024 Basic System Information: --------------------------------- Uptime : 7 days, 7 hours, 0 minutes Processor : AMD EPYC-Rome Processor CPU cores : 4 @ 1996.250 MHz AES-NI : ✔ Enabled VM-x/AMD-V : ❌ Disabled RAM : 7.7 GiB Swap : 4.0 GiB Disk : 503.8 GiB Distro : Debian GNU/Linux 12 (bookworm) Kernel : 6.1.0-25-amd64 VM Type : KVM IPv4/IPv6 : ✔ Online / ✔ Online fio Disk Speed Tests (Mixed R/W 50/50) (Partition /dev/vda3): --------------------------------- Block Size | 4k (IOPS) | 64k (IOPS) ------ | --- ---- | ---- ---- Read | 31.02 MB/s (7.7k) | 1.59 GB/s (24.9k) Write | 31.06 MB/s (7.7k) | 1.60 GB/s (25.0k) Total | 62.09 MB/s (15.5k) | 3.19 GB/s (49.9k) | | Block Size | 512k (IOPS) | 1m (IOPS) ------ | --- ---- | ---- ---- Read | 2.42 GB/s (4.7k) | 3.48 GB/s (3.4k) Write | 2.54 GB/s (4.9k) | 3.71 GB/s (3.6k) Total | 4.96 GB/s (9.7k) | 7.20 GB/s (7.0k) YABS completed in 40 sec
-
What kind of performance are you looking for? Throughput or IOPS?
IOPS, since on my RS2000 they are very sluggish even at around 30% cpu and mem usage. I suspect is all down to to many small read/writes to the disk.
-
The price difference is too little, I'm just going to go with the root server.
-
IOPS, since on my RS2000 they are very sluggish even at around 30% cpu and mem usage. I suspect is all down to to many small read/writes to the disk.
Historically the IOPS on VPS have been limited to 300. That wasn't the case on the RS however. I don't know if this is still the case with the NVMe drives.
-
Historically the IOPS on VPS have been limited to 300. That wasn't the case on the RS however. I don't know if this is still the case with the NVMe drives.
According to @tab's YABS above, it's much higher now.
-
According to @tab's YABS above, it's much higher now.
That's the burst performance over 30s. That's not the sustained performance.
-
That's the burst performance over 30s. That's not the sustained performance.
So how long should we test to get the sustained performance? 1h, 1d ...? It's still an 1h or 1d burst test then of course, but a longer one than 30s. I hope it works out better than it did at *VH in 2021.
-
I hope it works out better than it did at *VH in 2021.
That was a burn in test... - anyway thank you for your burning question.
I'd say about 30 minutes - if it doesn't catch fire until then, it should be good.
-