991 字
5 分钟
Netcup Germany RS 2000 G9.5 PRO Benchmark
Netcup’s June promo unexpectedly brought back the once red‑hot RS2000 series with more resources and a lower price. Before it returned to the official site, this plan was already being scalped on the market with a ~€300 premium. Hard to imagine how many people paid that markup only to see this offer come back. There’s also a lottery for the US region this time, with a chance to draw a G11 9634. For the German variant, IO performance is inconsistent: some instances have excellent IO, others are quite poor, likely due to dynamic allocation on nodes crowded with heavy users.
Registration guide: Netcup Tax-Free Registration Guide
The G9.5 bandwidth limiting policy that most people care about is basically the same as before, but it still needs to be verified by testing.
1. Guaranteed baseline bandwidthDefault bandwidth: Each Root Server has a guaranteed minimum bandwidth of 1 Gbps, even if its NIC is 2.5 Gbps (i.e., the theoretical maximum is 2.5 Gbps).
Shared utilization: As long as no throttling condition is triggered, the server can temporarily use more than 1 Gbps bandwidth (up to 2.5 Gbps), but network resources must be fairly shared with other users.
2. Conditions for triggering traffic and bandwidth limitsWhen the following two conditions are met at the same time, the server bandwidth will be limited to 200 Mbps:
Condition 1: Monthly total traffic exceeds the thresholdThreshold: Monthly traffic > 120 TB
Example calculation:At a constant 1 Gbps rate, theoretical monthly traffic ≈ 1 Gbps × 30 days ≈ 324 TB.The actual 120 TB threshold is about 37% of the theoretical value, indicating the policy allows relatively high usage but prohibits extreme, continuous consumption.
Condition 2: Sustained high bandwidth usageThreshold: Average bandwidth > 1 Gbps for more than 60 consecutive minutes
Monitoring logic:The system monitors using a sliding time window (e.g., sampled every minute). If the 60‑minute average continuously exceeds 1 Gbps, throttling is triggered.
Exceptions:Short bursts (e.g., saturating 2.5 Gbps for 10 minutes) will not trigger throttling because the time window requirement is not met.
3. Throttling measuresBandwidth cap: Once triggered, the server connection will be capped at 200 Mbps (20% of the original guaranteed value).
Purpose: To prevent a single user from occupying excessively high bandwidth for long periods and to ensure other customers’ network performance (fair‑use principle).
4. Key usage scenariosCompliant usage examples:
Average daily traffic ≤ 4 TB (120 TB/month), with no long‑term full‑bandwidth usage → 1 Gbps guarantee is maintained.
A burst at 2.5 Gbps for 30 minutes → No throttling (duration is below the 60‑minute threshold).
Non‑compliant usage examples:
Continuous 2 hours at 1.2 Gbps (average bandwidth above threshold + duration above threshold) → Throttled to 200 Mbps.
Monthly traffic reaches 150 TB (even with stable bandwidth usage) → Throttled (and Condition 2 must also be checked).Plan
-
Provider : Netcup
-
Type/Plan : RS 2000 G9.5 PRO
-
Processor : AMD EPYC 7702P 64-Core Processor
-
Num of Core : 6 Cores
-
Memory : 1****6 GB
-
Storage : 64****0 GB NVMe
-
Bandwidth : 120****TB @ 2.5 Gbps IN | 2.5 Gbps OUT
-
Location : DE
-
Price : €10.75
Benchmark
Yabs
Basic System Information:---------------------------------Uptime : 0 days, 7 hours, 42 minutesProcessor : AMD EPYC 7702P 64-Core ProcessorCPU cores : 6 @ 1996.246 MHzAES-NI : ✔ EnabledVM-x/AMD-V : ❌ DisabledRAM : 15.6 GiBSwap : 0.0 KiBDisk : 629.8 GiBDistro : Debian GNU/Linux 12 (bookworm)Kernel : 6.1.0-37-amd64VM Type : KVMIPv4/IPv6 : ✔ Online / ✔ Online
IPv6 Network Information:---------------------------------ISP : netcup GmbHASN : AS197540 netcup GmbHHost : NETCUP-GMBHLocation : Nuremberg, Bavaria (BY)Country : Germany
fio Disk Speed Tests (Mixed R/W 50/50) (Partition /dev/vda3):---------------------------------Block Size | 4k (IOPS) | 64k (IOPS) ------ | --- ---- | ---- ----Read | 142.24 MB/s (35.5k) | 1.40 GB/s (21.9k)Write | 142.61 MB/s (35.6k) | 1.40 GB/s (22.0k)Total | 284.86 MB/s (71.2k) | 2.81 GB/s (43.9k) | |Block Size | 512k (IOPS) | 1m (IOPS) ------ | --- ---- | ---- ----Read | 3.44 GB/s (6.7k) | 3.53 GB/s (3.4k)Write | 3.62 GB/s (7.0k) | 3.77 GB/s (3.6k)Total | 7.07 GB/s (13.8k) | 7.30 GB/s (7.1k)
iperf3 Network Speed Tests (IPv4):---------------------------------Provider | Location (Link) | Send Speed | Recv Speed | Ping----- | ----- | ---- | ---- | ----Clouvider | London, UK (10G) | 2.70 Gbits/sec | 2.30 Gbits/sec | 22.6 msEranium | Amsterdam, NL (100G) | 2.71 Gbits/sec | 2.32 Gbits/sec | 16.8 msUztelecom | Tashkent, UZ (10G) | 2.48 Gbits/sec | 1.41 Gbits/sec | --Leaseweb | Singapore, SG (10G) | 1.29 Gbits/sec | 1.55 Gbits/sec | --Clouvider | Los Angeles, CA, US (10G) | 1.08 Gbits/sec | 678 Mbits/sec | --Leaseweb | NYC, NY, US (10G) | 1.30 Gbits/sec | 1.82 Gbits/sec | 91.4 msEdgoo | Sao Paulo, BR (1G) | 2.00 Gbits/sec | 57.6 Mbits/sec | --
iperf3 Network Speed Tests (IPv6):---------------------------------Provider | Location (Link) | Send Speed | Recv Speed | Ping----- | ----- | ---- | ---- | ----Clouvider | London, UK (10G) | 2.56 Gbits/sec | 2.28 Gbits/sec | 22.5 msEranium | Amsterdam, NL (100G) | 2.65 Gbits/sec | 2.29 Gbits/sec | 16.7 msUztelecom | Tashkent, UZ (10G) | 2.43 Gbits/sec | 1.70 Gbits/sec | 90.6 msLeaseweb | Singapore, SG (10G) | 2.28 Gbits/sec | 1.93 Gbits/sec | 151 msClouvider | Los Angeles, CA, US (10G) | 955 Mbits/sec | 615 Mbits/sec | 156 msLeaseweb | NYC, NY, US (10G) | 2.49 Gbits/sec | 1.89 Gbits/sec | 91.6 msEdgoo | Sao Paulo, BR (1G) | 1.19 Gbits/sec | 32.5 Mbits/sec | 198 ms
Geekbench 6 BenchMark Test:---------------------------------Test | Value |Single Core | 1047Multi Core | 4386Full Test | https://browser.geekbench.com/v6/cpu/12586254
YABS completed in 14 min 33 secStreaming Benchmark
Netcup Germany RS 2000 G9.5 PRO Benchmark
https://catcat.blog/en/netcup-de-rs-2000-g9-5-pro-benchmark.html