all that has been addressed on this new servers setup so now instead of running pretty much just one main high-end server from a budget provider and a few other supporting (lowend/legacy) servers that can't really do much else I now run two actually capable mid-end servers at all times from direct subsidiaries of top-tier ISO 27001 certified providers along with a small supporting cloud based server ran by the operator of the previous main server datacenter that it's just enough to run this site and route some traffic all at the same combined 65 €/month budget.
let's get started with the now primary server card then:
- motehrboard: MSI m-ATX B350M PRO-VDH (MS-7A38) A.G5 (AMD Zen Summit Ridge Windows 10)
- chipset: AMD B350
- monitoring: NUVOTON 6795D Controller Chip
- vrm: Richtek RT8894A
- cpu: AMD AM4 Ryzen 7 PRO 1700X Processor (8x2x3.4-3.8 GHz 8x32 KiB L1d 8x64 KiB L1i 8x512 KiB L2 2x8 MiB L3 14nm 95W)
- memory: 2x16 GiB DDR4 2133MHz Crucial CT16G4DFD8213 (15-15-15-36)
- nic: Realtek 8111H Gigabit LAN controller (PCIe x1 2.5GT/s)
- ssd: Samsung SSD 850 EVO 250GB (SATA 3.1 6.0 Gb/s 2.5 inches 2M hours MTBF fw: EMT02B6Q)
- fan: 1600 rpm generic cpu fan
- system: up-to-date Fedora release 38 (Thirty Eight)
- kernel: up-to-date 64-bit Linux kernel 6.4.14 (microcode: 0x8001137)
- reboot: ~1.5 min (typical)
- condition: used hardware
- passmark: https://www.passmark.com/baselines/V11/ ... 4415408605
- CPU Score: 15,474
- Single Thread: 2,106 MOps/Sec
- RAM Score: 2,181
- Gaming Score: 2,005
- CPU Score: 15,474
- datacenter: myLoc managed IT AG (ISO 27001 Certified)
- location: Am Gatherhof 44, 40472 Düsseldorf, Germany
- network: 1000 Mbit Gamerflat (Hewlett Packard)
- hostname: fpsclasico.eu
- diagnosis: looking glass | smokeping
- speedtest: https://www.speedtest.net/result/c/a418 ... e6ebb0fe1c
- type of product: mid-end dedicated game server
- usage: currently decommissioned
- price: 32.02 €/Month
- regular hardware monitor sample
Code: Select all
23:20:23 up 17:18, 1 user, load average: 0.94, 0.98, 1.00
CPU:
core 0:
thread 0:
freq: 3399 MHz
thread 1:
freq: 3399 MHz
core 1:
thread 0:
freq: 3399 MHz
thread 1:
freq: 3399 MHz
core 2:
thread 0:
freq: 3399 MHz
thread 1:
freq: 3399 MHz
core 3:
thread 0:
freq: 3399 MHz
thread 1:
freq: 3399 MHz
core 4:
thread 0:
freq: 3399 MHz
thread 1:
freq: 3399 MHz
core 5:
thread 0:
freq: 3399 MHz
thread 1:
freq: 3399 MHz
core 6:
thread 0:
freq: 3399 MHz
thread 1:
freq: 3399 MHz
core 7:
thread 0:
freq: 3399 MHz
thread 1:
freq: 3399 MHz
freq: 3399.274 MHz
temp: 51.2℃
M/B:
temp (CPU): 48.0℃
temp (PCH): 56.0℃
temp (VRM): 63.0℃
temp (M/B): 42.0℃
fan (CPU): 1525 rpms
fan (M/B): 0 rpms
VCORE: 1.120 V
VRAM: 1.200 V
12VCC: 12.288 V
5VCC: 5.040 V
3VCC: 3.392 V
3VSB: 3.376 V
VBAT: 3.344 V
VRM:
VDD: 1.092 V
IDD: 43%
VNB: 0.749 V
INB: 22%
MEM:
bios: 736.12 MiB
used: 1096.32 MiB
SSD:
temp: 36.0℃
written: 1848.57 GiB
workload: 7.38%
health: PASSED
LAN:
multicast: 125193
rx_bytes: 5398635638
rx_packets: 64897321
tx_bytes: 9717984287
tx_packets: 23955366
as for the network is pretty much the same if not better setup. the offer itself is also a top-tier premium 1 gbit however the bandwidth seems to be (generally) much better on this spot than on the previous one where outside of the top destinations there was locations where the bandwidth was more limited from what I could test. I mean, that was a common (verified) complain on the previous setup ("slow" downloads) hopefully that seems improved now. the only real downsides of the offer (in comparison) are lack of any real IPv6 support and sub-par location but when you take everything in consideration those are kind of minor details. although not really supported IPv6 still works all the same and sure although the Frankfurt location could arguably be better (i.e. optimal) this Dusseldorf location is by no means any worse. I know for many Dusseldorf can probably look just like any other random spot lost in the map to run a server but truth to be told after operated the server the location is a lot more competitive and versatile that I would have anticipated. Dusseldorf sits itself as the main network center of the most populous state in Germany (North Rhine-Westphalia) and acts as a major hub between the two biggest network spots in Europe, Frankfurt and Amsterdam, with important international network points of presence as well. matter of facts, netcologne (that is a major German isp) clients are now pretty much on a lan with this server so as usual the final result will obviously vary for each specific connection but overall on average final in-game ping should be the same or just about the same (i.e. -/+ 1-3 msecs of difference) if not better specially considering the whole true multi homed/location server setup now.
let's move on to the new cloud based web services server card then:
- hardware: Supermicro AMD Epyc System (AMD SP3 Zen 3 Milan RHEL 8.3)
- cpu: 2x2.45-3.5 GHz dedicated threads (AMD AM2 Athlon 64 X2 Windsor 6400+ BE compatible emulated)
- memory: 4 GiB Samsung DDR4 RegECC (3200MHz)
- storage: 40 GiB Enterprise NVMe PCIe 3.0 (Ceph Cluster)
- system: up-to-date Fedora Linux 39 (Cloud Edition)
- kernel: up-to-date 64-bit Linux kernel 6.9.6 (microcode: 0x1000065)
- reboot: ~30 sec (typical)
- monitor: mrtg.fpsclasico.de
- condition: virtualized hardware
- passmark: https://www.passmark.com/baselines/V11/ ... 6843912518
- CPU Score: 2,561
- Single Thread: 2,227 MOps/Sec
- RAM Score: 1,769
- Gaming Score: 3,660
- CPU Score: 2,561
- datacenter: interwerk (green electricity -250Gbit)
- location: Dieselstraße 37, 60314 Frankfurt am Main, Germany
- network: Shared 20 Gbit Optic Cluster, 200 Mbit Fair-Use (Juniper Networks)
- hostname: fpsclasico.de
- diagnosis: looking glass | smokeping
- speedtest: https://www.speedtest.net/result/c/da0d ... 312ed94c0b
- type of product: entry-level "dedicated" (resources guaranteed) cloud root server
- usage: currently decommissioned
- price: 4.90 €/Month
as for the network it's limited to 200 mbps still that's more than enough for the use being of serving this site and routing game server traffic. the location is probably the flagship feature of this product, as hinted before this is run exactly on the same site as the previous main server (i.e. interwerk, Frankfurt) only that this time instead of being provided by a reseller directly by the own datacenter operators themselves so even in the event somebody getting sub-par ping at some of the new servers they can always just connect through this route to guarantee a sensible latency in tune with the previous one. on the downside being smaller this provider probably won't match the likes of amazon, google, azure or just of the two other top-tier used providers in terms of reliability so apparently they don't count with a ISO 27001 certification for guarantee (to a high degree) continuity of service still again as some datacenter operators is a important improvement on that aspect over the previous main provider were discontinuity of service (at some point) was basically expected instead.
last but not least let's finish with the new secondary server card then:
- motherboard: FUJITSU μBTX D2981-A1 6.00 R1.11.2981.A1 (AMD K10 Thuban Windows 7)
- northbridge: AMD 785G (RS880) [HyperTransport 3.0 2 GHz and PCI Express 2.0]
- southbridge: AMD SB710 [4xSATA II RAID]
- monitoring: Microchip SCH5627 SuperI/O
- cpu: AMD AM3 Phenom II X6 1055T (6x2.8-3.3 GHz 6x64 KiB L1d 6x64 KiB L1i 6x512 KiB L2 6 MiB L3 45nm 95W)
- igpu: ATI Radeon HD 4200 010.094.001.047.000000 (40:4:4 2 CU 500 MHz 256 MiB VRAM 256 MiB Shared Memory 55nm 15W)
- memory: 4x4 GiB DDR3 1333MHz Samsung M378B5273DH0-CH9 (256Mx8 DR 9-9-9-24)
- nic: Broadcom Inc. and subsidiaries NetLink BCM57780 Gigabit Ethernet (PCIe x1 2.5GT/s)
- audio: Realtek ALC663 (5.1 High Definition Audio)
- hdd1: 2 TB WD Gold Enterprise WD2005FBYZ-01YCBB3 (SATA 3.1 6.0 Gb/s 3.5 inches 7200 rpm 128 MB 2.5M hours MTBF fw: RR09)
- hdd2: 2 TB Seagate Exos 7E8 ST2000NM0055-1V4104 (SATA 3.1 6.0 Gb/s 3.5 inches 7200 rpm 128 MB 2M hours MTBF fw: TN05)
- fans: high performance proprietary 13k rpm cpu and 9k rpm system fans
- system: up-to-date Fedora Linux 40 (Workstation Edition)
- kernel: up-to-date 64-bit Linux kernel 6.10.9 (microcode: 0x10000dc)
- reboot: ~1 min (typical)
- monitor: mrtg.fpsclassico.com
- condition: end-of-life used hardware
- passmark: https://www.passmark.com/baselines/V11/ ... 7303320422
- CPU Score: 3,665
- Single Thread: 1,468 MOps/Sec
- RAM Score: 1,286
- Gaming Score: 885
- CPU Score: 3,665
- datacenter: United Internet AG (ISO 27001 TÜV-certified, ISO 50001 green electricity)
- location: Brauerstr. 48, 76135 Karlsruhe, Germany
- network: 100 Mbit Unmetered (Cisco Systems, Inc)
- hostname: fpsclassico.com
- diagnosis: looking glass | smokeping
- speedtest:
- type of product: legacy mid-end small business general purpose dedicated server
- usage: web services server, historic competitive stats server, files and video serving, video production and dedicated workstation, data mirroring
- price: 28.47 €/Month [7.12 €/Month - 12 Months off]
- hardware monitor sample while producing video
Code: Select all
17:21:39 up 1 day, 11:18, 2 users, load average: 4.95, 5.18, 4.35
CPU:
core 0:
freq: 2793 MHz
core 1:
freq: 2793 MHz
core 2:
freq: 2793 MHz
core 3:
freq: 2793 MHz
core 4:
freq: 2793 MHz
core 5:
freq: 2793 MHz
freq: 2793.167 MHz
temp: 58.1℃
GPU:
freq: 494.04 MHz
M/B:
temp (CPU): 58.0℃
temp (GPU): 77.8℃
temp (N/B): 41.2℃
temp (S/B): 39.3℃
fan (CPU): 7595 rpms
fan (M/B): 8232 rpms
VCPU: 1.120 V
VRAM: 1.498 V
3VCC: 3.301 V
3VSB: 3.288 V
VBAT: 3.000 V
MEM:
bios: 659.02 MiB
used: 1181.43 MiB
Status: No errors detected
HDD1:
temp: 35.0℃
written: 1.07 TiB
read: 0.02 TiB
workload: 66.80%
rate: 149.97 MiB/s
health: PASSED
HDD2:
temp: 38.0℃
written: 84.61 TiB
read: 71.81 TiB
workload: 3.49%
rate: 155.86 MiB/s
health: PASSED
LAN:
multicast: 98156
rx_bytes: 6.108 GB
rx_packets: 61.784 Mp
tx_bytes: 7.120 GB
tx_packets: 23.931 Mp
another key part of this server is the legacy low-end integrated gpu on the northbridge. sure for other than regular office use and running old games like these here under standard hd low quality settings that igpu is pretty much useless by today standards (I mean, it's not even a old agp card, let alone a basic pcie one, but really a legacy pci one instead) but the thing is that's exactly just what's needed here. on this aspect the old main server clearly outperformed this secondary one by a huge margin at least under the normal usual settings however just after a few adjustments and optimizations it ended up performing all the same if not better even on that aspect. overall, the machine is just capable of running Quake III Arena under the default UnFreeZe hd settings at 85 fps or at 25 fps for the 2k mp2/mp4 video renders converted at around 1:5.5-1:6.5 real-time ratios all the same which is just what's needed. another strong point here is the provider, this comes just from the business servers branch of the same group as the well known international internet provider 1&1 so the quality should come with it. I mean, I just ordered this a Saturday eight in the morning, they (over)delivered it by nine in the morning and by nine and a half in the morning the server was up and running the emergency local physical backup as-is, no bullshit. on the downside you have the network, the port is capped just at 100 mbit (though unmetered at least) as back in the day still that's more than enough for the use that it's gonna have the server and the location (Berlin) it might not be as competitive though I must also say that even on that aspect it impressed me and the latency there it's a lot better than anticipated. I mean, you have people playing at the secondary UnFreeZe server on a daily basis with the same or almost same ping as on any of the other servers if not better. obviously at that price tag you can't expect to have it all and main it still this is just the perfect fit for a reliable secondary and backup server to use.
for the system I deployed all the same up-to-date images on all of them with no virtualization involved whatsoever this time so I assume that unless something breaks these should go at least for another couple of years. and well I think that this should be all about these new fpsclassico 2023 servers report at least by now so now you know what has been done with your funds. hopefully the only good about the failure was the timing that happened just one week before renewal which means that only 1 week worth of service (the week of failure around 15 euros) has been lost beside a extra month I took from the old legacy server (i.e. 6 euros) to keep the data redundant some more time just to be safe so virtually no money got lost (21 euros is typically not even the cost of delivering just one server to put it in perspective) and the total costs of the new servers remained pretty much the same (1.84 euros less per month in total in fact), no big deal. matter of facts, after have finally used the new servers for some time now I can tell that when it comes to economy and performance per cost metrics these new servers clearly outperform there the previous ones by a big margin (+65% improvement) in that aspect so they are clearly a much better deal in comparison.