All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
fastpipe.io - Cloud Servers (KVM) - Frankfurt, Germany - NVMe Cached - starting at 3,00€!
jh_aurologic
Member, Patron Provider
fastpipe.io is a brand of combahton GmbH, offering Cloud and Dedicated Servers in Frankfurt am Main, Germany.
Our services are based on latest HPE and Supermicro servers, offering outstanding realiability and performance. We also operate our own fully redundant, multi-datacenter Juniper based network, consiting of Juniper EX, QFX and MX equipment.
Some key features of our SSD Cloud Servers
High Availability: All SSD Cloud Servers are provisioned on a redundant, highly available KVM Cluster consisting of multiple Cloud and Storage Nodes. HA is provided by Corosync and GlusterFS. Storage Nodes are based on SSD and HDD (SAS) Raid-10, using NVMe Drives (Caching) to provide exceptionel I/O performance. The whole setup is fully redundant, both on network and server level.
Excellent connectivity: By default, every cloud node is connected with 2x 10Gbit to our network. This provides network-level fault resilience, fast data transfers, and best performance even at peak times. Network-intensive applications are excellently hosted with us, as our AS30823 is based on a premium Tier-2 blend. Packetloss during peak hours? Not when your server is hosted with us.
DDoS-Protection: Our cloud servers include our exclusive, self developed DDoS protection, which is unique beside other proprietary solutions and used by large gameserver clients. Even DDoS attacks on web applications are filtered out by our DDoS protection.
Additional IP-Addresses: You can add additional IP addresses (from 1.49€) and subnets (from 6.95€) to your instance. The setup takes place immediately, the reserved address ranges are routed to an instance of your choice.
Daily Backup: Data loss is annoying. Accordingly, our cloud server nodes are equipped with several hard drives (hardware raid) as standard. You can avoid the risk of self-generated data loss with minimal effort through our automated snapshot backups. For this purpose we make a complete backup of your server every day.
Scale at any time: Do you need more CPU power, memory or disk space instantly? Scale your server automatically to the next higher level, fully automated and without changing the existing configuration or data. We only charge you the difference in connection with the current remaining term.
Virtual Private Lan: Create your own, isolated networks with our Virtual Private Lan (VPL). VPLs behave like V-Lans and provide an internal network between different instances where you can freely use your own IP addresses. Choose between 1Gbit, 2Gbit or 5Gbit interfaces.
Free BGP (AS30823): Already own an AS and IP-Space? Just open a ticket after your server has been provisioned and tell us your ASN and Prefixes with valid route objects. We will configure the session within two business days.
Real hardware required?: Check out our website, beside Cloud Servers, we're also offering Dedicated Servers based on premium hardware.
Small entry offerings
With a small budget, you get a lot of power and possibilities with our Cloud Servers. Starting at 2,95€ ($3,56) per 30 days, we offer:
- Cloud HE-2
- One CPU Core (Intel Xeon E5v2/v3/v4 @ +2,5GHz) incl. AES-NI
- 2048MB ECC Memory
- 80GB HDD Storage
- Debian, Ubuntu, CentOS and FreeBSD available as one-click installation
- Additional details and order link available on our website
More power needed? Choose a higher package or upgrade with just a few clicks, without moving configuration or data.
Want to learn more about our current progress? Checkout our blog.
Comments
Does have VPS instance 10 Gbit shared uplink, or is limited/capped to 1Gbit?
and Anyone who does have any longer experiences with this provider, or any benchmarks available to share here, pls? thanks
Pretty sure if you spend some time going through their past discussions or even search within LET you should be able to find reviews/feedback and benchmarks
Every node has 2x 10G, you can expect more than 1G (guaranted), however not full 10G. For full 10G, you better get a dedicated server.
@miu: The usage of the search function is not a crime.
Might not be comparable to nowadays performance, we moved almost all nodes to our new cloud infrastructure over the past months, which is by far faster and more reliable than single nodes.
Yeah I'm aware of the changes based on your previous posts/offers, but he said "any benchmarks available" - hence my response
There should be plenty :-)
To be honest, this is a lot better then what it was about three months ago. It was basically unusable back then, even the ones advertised as SSD.
@berkay Thank you for share!
It seems they have 2 x 10G for both network and storage/disk I/O.
Maybe a proper network test would help clarify, if any new subscriber would care to oblige...?
curl -LsO bench.monster/speedtest.sh; bash speedtest.shThat’s not really a proper network test though... Speedtest.net is woefully overrated and very inconsistent and unreliable on higher port speeds.
No network test out there could be deemed perfect, but at least it's a better indicator than many.
Nah, just run iperf. Or the built-in iperf in Yabs. Speedtest.net really I only fit for home connections <200 Mbps.
"""SSD"""
Clearly listed as HDD in the offer?
Agreed in principle, but how many times do you need to ping Paris for an offer based in erm Paris (or near enough)...?
And then comes...
Bogor, Indonesia Moscow, RUNot the same storage, not the same offer as within this thread; totally different volumes and no nvme cache here, comparing apples with bananas on the storage layer, however a look on our status page might be worthful: https://status.combahton.net/#scheduled-142
Feel free to contact our customer service in order to get your servers disk migrated on another volume, it even doesnt require a shutdown - it‘s done as live migration and will turn your benchmark by 180 degrees (if your servers i/o is not already limited due to excessive use)!
Doesnt matter - we run a few more storage nodes than just one for SSD - recent HDD storages will even outperform old SSD ones, we easily got write speeds of 2GB/s on HDD when we installed four around 3 weeks ago - together with NVMe bcache its lightning fast
Bad performance can have multiple reasons, excessive I/O abuse or some operation on the storage layer (customer service got you covered).
To clarify this a bit, beside some other el cheapo companies on LET, we have four storage servers (storage nodes) per glusterfs volume (distributed+replicated) by default. Each volume is a lot more failsafe than a single server with raid due to arbiters and replication (no spof).
A glusterfs volume can be built either on HDD, SSD or NVMe.
All recent storage nodes got each at least 2x 10Gbit to 2x Juniper QFX switches (redundantly as Virtual Chassis) and at least 2x 500GB NVMe Cache using PCIe NVMe (Raid 1, bcache) - drives itself using Raid 10. Nothing is stored on the Cloud Nodes, where Cloud Servers run on. Also no storage traffic passes over the external network, thats all isolated down to Layer 1, completely seperated from our core network.
no very impressive results for 10Gbit..
Carry out the same test with more cpu ressources and not the smallest instance we offer and it will be a lot more, quite simple
To be fair, the smallest instance is exactly what you're advertising here, so it's reasonable that people would be benchmarking it.
Of course - but you cant expect multiple gigabit from a small virtual machine, network throughput depends heavily on cpu performance. Small instance and high network throughput will be not possible due to the nature of how networking is done within a vm.
We didnt advertise, that anyone will be able to push 10G through our smallest instance, so I guess it's kinda fair to test with a larger instance using proper methods (iperf), instead of just claiming network quality is overall not to good
OK, based on your latest claims, here is it - my 1vcore smallest 512RAM instance for $20/yearly by HostUS. They are then probably from other planet in accordance what u wrote.
BTW: i think that CPU is not bottleneck in mentioned case above, because i have bunch of 1Gbit VPSes also with single vcore (and similar CPU) that can beat without problem your 10G
explain next.. regards
OS : CentOS 7.8.2003 (64 Bit)
Virt/Kernel : KVM / 3.10.0-1127.el7.x86_64
CPU Model : Intel(R) Xeon(R) E-2276G CPU @ 3.80GHz
CPU Cores : 1 @ 3791.994 MHz x86_64 16384 KB Cache
CPU Flags : AES-NI Enabled & VM-x/AMD-V Enabled
Load Average : 0.11, 0.04, 0.02
Total Space : 15G (986M ~7% used)
Total RAM : 486 MB (91 MB + 309 MB Buff in use)
Total SWAP : 0 MB (0 MB in use)
Uptime : 0 days 0:1
ASN & ISP : AS7489, HOSTUS-AMS01
Organization :
Location : Amsterdam, Netherlands / NL
Region : North Holland
## Global Speedtest
Location Upload Download Ping
Nearby 2845.36 Mbit/s 250.59 Mbit/s 2.958 ms
USA, New York (Sprint) 229.45 Mbit/s 214.84 Mbit/s 85.997 ms
USA, Chicago (Windstream) 215.81 Mbit/s 310.31 Mbit/s 94.277 ms
USA, Dallas (Frontier) 165.61 Mbit/s 361.33 Mbit/s 119.285 ms
USA, Miami (Frontier) 178.81 Mbit/s 384.42 Mbit/s 109.652 ms
USA, Los Angeles (Spectrum) 138.85 Mbit/s 319.53 Mbit/s 134.734 ms
UK, London (toob Ltd) 273.68 Mbit/s 743.15 Mbit/s 8.275 ms
France, Lyon (SFR) 287.48 Mbit/s 1259.64 Mbit/s 14.361 ms
Germany, Berlin (DNS:NET) 1312.80 Mbit/s 2862.37 Mbit/s 13.689 ms
Spain, Madrid (MasMovil) 450.15 Mbit/s 430.50 Mbit/s 35.176 ms
Italy, Rome (Unidata) 591.17 Mbit/s 501.59 Mbit/s 32.806 ms
Russia, Moscow (MTS) 70.51 Mbit/s 756.24 Mbit/s 47.258 ms
Israel, Haifa (013Netvision) 226.48 Mbit/s 81.32 Mbit/s 79.337 ms
India, New Delhi (GIGATEL) 111.29 Mbit/s 147.32 Mbit/s 149.104 ms
Singapore (FirstMedia) 57.37 Mbit/s 18.55 Mbit/s 184.759 ms
Japan, Tsukuba (SoftEther) 14.00 Mbit/s 32.80 Mbit/s 259.928 ms
Australia, Sydney (Optus) 4.52 Mbit/s 11.71 Mbit/s 371.593 ms
RSA, Randburg (Cool Ideas) 110.21 Mbit/s 156.96 Mbit/s 172.332 ms
Brazil, Sao Paulo (Criare) 53.00 Mbit/s 116.06 Mbit/s 220.237 ms
## ## ## ## ## ## ## ## ## ## ## ## ## ##
Yet-Another-Bench-Script
v2020-06-20
https://github.com/masonr/yet-another-bench-script
## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ##
Tue Jul 21 19:59:55 UTC 2020
Basic System Information:
Processor : Intel(R) Xeon(R) E-2276G CPU @ 3.80GHz
CPU cores : 1 @ 3791.994 MHz
AES-NI : ? Enabled
VM-x/AMD-V : ? Enabled
RAM : 486M
Swap : 0B
Disk : 15G
fio Disk Speed Tests (Mixed R/W 50/50):
iperf3 Network Speed Tests (IPv4):
Provider | Location (Link) | Send Speed | Recv Speed
| | |
Bouygues Telecom | Paris, FR (10G) | 6.03 Gbits/sec | 8.57 Gbits/sec
Online.net | Paris, FR (10G) | 8.58 Gbits/sec | 5.20 Gbits/sec
WorldStream | The Netherlands (10G) | 8.49 Gbits/sec | 8.58 Gbits/sec
wilhelm.tel | Hamburg, DE (10G) | 8.31 Gbits/sec | 8.47 Gbits/sec
Biznet | Bogor, Indonesia (1G) | 634 Mbits/sec | 313 Mbits/sec
Hostkey | Moscow, RU (1G) | 918 Mbits/sec | 686 Mbits/sec
Velocity Online | Tallahassee, FL, US (10G) | 1.07 Gbits/sec | 1.02 Gbits/sec
Airstream Communications | Eau Claire, WI, US (10G) | 1.01 Gbits/sec | 173 Mbits/sec
Hurricane Electric | Fremont, CA, US (10G) | 972 Mbits/sec | 331 Mbits/sec
BTW2: another thing is that what u offer for its price is good offer imo (KVM w 80GB storage and 2GB RAM for 3€ monthly is good price imo). But what i suggested is that network speed results would be better on 10G. That's all, Nothing more against you and good luck with sale
They are generally aggressive, nothing special against you.
And they aren't very helpful or friendly when it comes to their support either.
The pricing and resources are good. But it would be reassuring to have the smallest instances tuned to guarantee 1gbps for bursts.