All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
RackNerd BF VPS benchmark and review
OK, OK, I get it, how can a LET benchmarker not have benchmarked and reviewed some RackNerd product? After all, @dustinc / RackNerd is one of the most prominent providers here on LET and seems to have a huge fan base. My excuse: Although I hinted a few times at dustinc that I'd be happy to test one of his products I never got a reaction from him, so I purchased one of this years BF VPS, a 1 core, 2.5 GB memory, 40 GB NVMe, 3 TB or so (doubled meanwhile because I joined the mindless "comment your order number to get double bandwidth" game - which I count as a BIG MINUS for every provider who plays that despicable game) traffic volume in Dublin, Ireland.
A small explanatory prolog seems to be required re "dustinc the accused 'criminal'" yada yada:
I know next to nothing about what really happened and am only interested in one single question: did a provider or did he not cheat or scam customers? And that was not even an accusation afaik. So for me it's simple, we are neither judges nor priests that is we are not tasked to judge other people. Would one like to be close friends with dustin may be an interesting question but one that is utterly irrelevant here in the context of LET.
Short version: I'm neither pro nor anti dustinc or RackNerd and here on LET as a reviewer/benchmarker I ONLY care about performance, bang per buck, reliability and customer service, period.
First impressions: RN offers an explicit choice between one off payment and subscription with PayPal. Kudos for that, I really appreciate that!
The RN panel is, well (yawn) a panel. It works, panel review done. But there is a 'but' with their VNC. It was s.n.a.i.l-slow when I used it to install and, even worse, only one keyboard layout is available. That's not acceptable for a provider operating and selling internationally.
Meanwhile dustinc (in his promo thread) told me that he "played a bit" with the VNC in Dublin and found it not at all slow. I don't care; when I had to use it it was cruelly slow. But it should be positively noted that dustinc right away looked at it, which means, he seems to actually care about customer feedback which is a big plus in my books. I'm in fact hopeful that they will offer additional keyboard layouts too.
Now to the "meat", to the benchmark, based on 60+ runs. As usual first processor and memory
Version 2.5.0a, (c) 2018+ jsg (->lowendtalk.com)
Machine: amd64, Arch.: amd64, Model: Intel(R) Xeon(R) CPU E5-2690 v4 @ 2.60GHz
OS, version: FreeBSD 14.0, Mem.: 2.479 GB
CPU - Cores: 2, Family/Model/Stepping: 6/79/1
Cache: 32K/32K L1d/L1i, 2M L2, 16M L3
Std. Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat
pse36 cflsh mmx fxsr sse sse2 ss sse3 pclmulqdq vmx ssse3 fma cx16
pdcm pcid sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline aes xsave
osxsave avx f16c rdrnd hypervisor
Ext. Flags: fsgsbase tsc_adjust bmi1 hle avx2 smep bmi2 erms invpcid rtm rdseed
adx smap umip syscall nx pdpe1gb rdtscp lm lahf_lm lzcnt
AES? Yes
Nested Virt.? Yes
HW RNG? Yes
ProcMem SC [MB/s]: avg 249.2 - min 99.3 (39.8 %), max 395.8 (158.8 %)
ProcMem MA [MB/s]: avg 557.2 - min 473.6 (85.0 %), max 705.0 (126.5 %)
ProcMem MB [MB/s]: avg 570.4 - min 462.2 (81.0 %), max 739.4 (129.6 %)
ProcMem AES [MB/s]: avg 734.6 - min 706.2 (96.1 %), max 741.3 (100.9 %)
ProcMem RSA [kp/s]: avg 105.8 - min 99.3 (93.8 %), max 110.3 (104.2 %)
Yay, me happily purring! Yes, it's "only" an E5v4 and certainly not a speed demon. But for a cheap VPS for less than $30/yr I'm pleased, usually E5v4 based (cheap) VPS I benchmark are around the 200 mark (PM-SC). crypto is known not to be a strong side of those processors but slightly over 100 RSA keypairs per seconds isn't bad at all and should be plenty enough for most web servers.
Now, on to the disk
--- Disk 4 KB - Buffered ---
Write seq. [MB/s]: avg 9.66 - min 9.06 (93.8%), max 10.21 (105.7%)
Write rnd. [MB/s]: avg 9.58 - min 9.42 (98.3%), max 9.98 (104.2%)
Read seq. [MB/s]: avg 15.86 - min 15.28 (96.3%), max 16.81 (106.0%)
Read rnd. [MB/s]: avg 13.97 - min 13.46 (96.4%), max 14.95 (107.0%)
--- Disk 4 KB - Sync/Direct ---
Write seq. [MB/s]: avg 9.53 - min 9.17 (96.2%), max 9.79 (102.7%)
Write rnd. [MB/s]: avg 9.46 - min 9.21 (97.4%), max 9.63 (101.8%)
Read seq. [MB/s]: avg 15.84 - min 15.22 (96.1%), max 18.15 (114.6%)
Read rnd. [MB/s]: avg 13.96 - min 13.38 (95.9%), max 15.07 (108.0%)
--- Disk 64 KB - Buffered ---
Write seq. [MB/s]: avg 113.15 - min 107.52 (95.0%), max 118.06 (104.3%)
Write rnd. [MB/s]: avg 125.62 - min 122.40 (97.4%), max 129.94 (103.4%)
Read seq. [MB/s]: avg 743.02 - min 568.56 (76.5%), max 1103.24 (148.5%)
Read rnd. [MB/s]: avg 192.85 - min 181.92 (94.3%), max 233.07 (120.9%)
--- Disk 64 KB - Sync/Direct ---
Write seq. [MB/s]: avg 15.02 - min 14.26 (94.9%), max 16.10 (107.2%)
Write rnd. [MB/s]: avg 7.42 - min 7.04 (94.9%), max 8.14 (109.7%)
Read seq. [MB/s]: avg 587.60 - min 564.12 (96.0%), max 675.49 (115.0%)
Read rnd. [MB/s]: avg 192.26 - min 183.46 (95.4%), max 222.49 (115.7%)
--- Disk 1 MB - Buffered ---
Write seq. [MB/s]: avg 179.32 - min 170.84 (95.3%), max 185.00 (103.2%)
Write rnd. [MB/s]: avg 379.08 - min 310.31 (81.9%), max 395.11 (104.2%)
Read seq. [MB/s]: avg 1936.40 - min 1538.21 (79.4%), max 2433.53 (125.7%)
Read rnd. [MB/s]: avg 938.91 - min 814.42 (86.7%), max 1047.87 (111.6%)
--- Disk 1 MB - Sync/Direct ---
Write seq. [MB/s]: avg 36.33 - min 34.08 (93.8%), max 42.59 (117.2%)
Write rnd. [MB/s]: avg 30.22 - min 29.02 (96.0%), max 34.72 (114.9%)
Read seq. [MB/s]: avg 1711.70 - min 1610.85 (94.1%), max 1791.51 (104.7%)
Read rnd. [MB/s]: avg 1016.63 - min 973.89 (95.8%), max 1082.43 (106.5%)
--- Disk IOps (Sync/Direct) ---
Write seq. **[MB/s]: avg 44.66** - min 41.92 (93.9%), max 46.11 (103.2%)
**IOps : avg 11432.86** - min 10731.43 (93.9%), max 11804.69 (103.3%)
WUT?? over 40 MB per second and solidly over 11k IOps?! That's insane. I think, dustinc should be put in front of the inquisition for sorcery! Excuse me sir, I buy a cheap BF VPS and you throw that kind of disk performance at me? And to add insult to injury you even keep the spread quite low, tssk.
As punishment I'll run my small DB benchmark, too. That'll teach you to stay away from witchcraft! (And all RackNerd haters to modestly shut up ...)
Comments
Now, lets look at the network, maybe we are luckier there. After all a cheap BF VPS must have a weak spot ...
Order: Europe, Asia, Americas, Oceanian & Africa
First, let me state front-up that I manually checked each and every target. All of them work. But it's by no means uncommon that one provider or ISP does reach some target fine while another doesn't.
I should also note that I put in one or a few "traps" that is, targets I know to be (tr)icky, otenet.gr being an example (which btw. my RN VPS did quite OK on).
For reasons unknown to me some @Clouvider targets, especially us-american ones, are weird, sometimes they do great, often they don't and sometimes they fail completely (Note: That doesn't mean that Clouvider POPs are crappy; there are as many potential culprits as there are hops to a target).
Another, relevant though, it seems to me side note: RackNerds network is part of the colocrossing and/or HostPapa AS, which is not bad at first glance (e.g. Telia/1299 is a major pipe of theirs) but seems to be ill reputed for reasons unknown to me. And WTF is "HostPapa"?
Well, whatever, connectivity to (or in my case within) Europe is decent; nothing to write home about but neither to complain about when considering the price of my BF VPS.
Connectivity to Asia is actually quite decent as well. Not so with the Americas; the east coast is plain crappy, the west coast seems to be somewhat of a lottery ranging from "BR, SAO is better (than LAX)" to San Francisco is quite OK. And Dallas is even great. Weird.
Finally Africa and Ozzyland. Connectivity to the latter is actually relatively decent (emphasis on 'relative') and with Africa it seems one must be happy to get there at all and with some luck with a Ozyyland-like performance. So, I guess, test passed.
All in all I wouldn't go so far to call connectivity the weak point of my RN BF VPS but it's certainly nothing to brag about. I think the label "well within the normal range for very cheap VPSs" would fit quite well.
TL;DR: A great choice if your focus is on disk-heavy loads like DBs and, provided you buy a VPS with at least two vCores, even worthy of a recommendation - with a caveat though; your use case should not require fast global connectivity.
Ping and web ping. Provide explanation.
colocrossing are considered lenient on abuse by some
their IP's are also on some blacklists which sucks if you wanna run email or vpn
(a) I don't accept commands from you (ask politely though and I'll explain it once more)
(b) I already did multiple times.
When you read you apply your pow. Apply more positive emotions. Since you write wall of text - don't be surprised that 95% of readers read only 5% of it. I can't find explanation on ping and web ping. Please.
I've had decent experiences with RackNerd, no complaints. The only real caveat for me is the CC ip ranges being on so many blocklists so less useful for some purposes.
ping is just that, a ping. Web ping tells how long it took to get a first reaction (i.e. not content but any reaction) from the target server. I came to do that because (a long time ago) I noticed that targets that actually were fast looked slow due to a "sleepy" http server. With a good, fast (in every sense) target server wep ping should be only minimally slower than std. ping.
Hm, web ping sounds like standard TTFB metric. Suggestion: be more clear on terminology by rethinking web ping term. It is confusing
I agree, disk is decent but on mine at least the CPU benchmark is pretty poor and no ipv6 in Ashburn. I picked something else up during the black friday sale and I'll be retiring my racknerd vps.
Wed Dec 4 08:18:12 AM EST 2024
Basic System Information:
Uptime : 8 days, 10 hours, 37 minutes
Processor : Intel(R) Xeon(R) CPU E5-2680 v2 @ 2.80GHz
CPU cores : 3 @ 2799.998 MHz
AES-NI : ✔ Enabled
VM-x/AMD-V : ❌ Disabled
RAM : 2.4 GiB
Swap : 2.5 GiB
Disk : 36.8 GiB
Distro : Debian GNU/Linux 12 (bookworm)
Kernel : 6.1.0-28-amd64
VM Type : KVM
IPv4/IPv6 : ✔ Online / ❌ Offline
IPv4 Network Information:
ISP : HostPapa
ASN : AS36352 HostPapa
Host : RackNerd LLC
Location : Ashburn, Virginia (VA)
Country : United States
fio Disk Speed Tests (Mixed R/W 50/50) (Partition /dev/vda1):
iperf3 Network Speed Tests (IPv4):
Geekbench 6 Benchmark Test:
Test | Value
|
Single Core | 343
Multi Core | 844
Full Test | https://browser.geekbench.com/v6/cpu/9238800
YABS completed in 24 min 21 sec
I don't comment on yabs, geekbench etc., but I'm not surprised at all, even ignoring the data you posted. Your VPS is based on an E5v2.
And btw, NO, disk is not "decent", it's amazingly great for a cheap BF VPS.
nah, it's still just decent. I've got 6 or 8 VPS from LET providers and frankly the only thing slower than the RackNerd disk is the MassiveGrid disk lol.
Here's my crunchbits disk, at the same yearly price, for comparison:
fio Disk Speed Tests (Mixed R/W 50/50) (Partition /dev/sda3):
Well comparing results from different benchmarks makes little sense. Besides you don't even show IOps (which is not "MB" or GB" but a simple number.
Based on the result sets I have, and that's quite a lot, your statement "frankly the only thing slower than the RackNerd disk is the MassiveGrid disk lol." is simply, pardon me, nonsense. I've seen plenty slower disks of all kind, incl. NVMe, from diverse providers, ranging from cheap Joe cheap VPS to not cheap but really nice VPS from reputable providers. The MG disk is not even in the same galaxy.
If you want to pick on RN, the NVMe is the worst choice because it's in the top 20% (prob. even 10%) I've seen so far.
But yabs does that (in brackets ()) next to the speed, which can be verified by multiplying block size * IOPS
@lothos said:
Block Size 4k (IOPS) 64k (IOPS)
Read 131.66 MB/s (32.9k) 1.12 GB/s (17.6k)
Write 132.01 MB/s (33.0k) 1.13 GB/s (17.7k)
Total 263.67 MB/s (65.9k) 2.26 GB/s (35.3k)
EDIT: better explanation and correct quoting
The IOPS is clearly marked as the number in parenthesis in my benchmarks.
I'm not picking on anything, I said it was decent. But in my result set of different LET VPS, the Racknerd disk is the second slowest, and it's in the bottom 20% for sure.
Hi @jsg -- Thank You for taking the time to write this detailed review! The performance numbers you've achieved are awesome to see.
Our new Dublin location has been incredibly popular since launching a couple weeks back. The demand we've seen validates our decision to expand there, and we're committed to growing our footprint in this region. Your benchmark results and real-world feedback help confirm we're delivering the performance our customers expect.
Thank You again for putting our services through their paces and for sharing these results with the community!
If there's anything you ever need, feel free to reach out to me directly at [email protected] or open a support ticket, we're truly available 24x7 to help.
P.S. By the way, we've submitted a feedback request for SolusVM to add more keyboard layouts for VNC, and I'm hopeful this will be available in future releases. Though as you pointed out earlier, VNC is more of something used once and forgotten about typically - i.e. for OS installations, firewall lockouts, etc. - for those reading, SSH remains the recommended way for daily server management
Hi @zedzed -- Thank You for sharing your experience with RackNerd. Really glad to hear that your experience has been overall solid! In terms of IP reputation, I totally understand what you're saying. At times, reputation can vary and unfortunately that's something any provider in this industry can experience, especially with larger IP ranges. The good news though is that these situations tend to be temporary in most cases, and can be resolved.
We take IP reputation seriously and work hard to minimize abuse that are within our control. Sometimes these situations can even stem from upstream customers who aren't even our direct clients i.e. if the blacklisting is due to a larger parent range or ASN level. A lot depends on which third party reputation service you're checking against as well.
If you ever need anything specific, don't hesitate to reach out and we'll be happy to help
We do have other upstreams and locations available as well.
Howdy @lothos -- we sincerely value your business and understand that your current needs have changed from your original deployment needs - and the good news is that we can custom tailor solutions that meet your today needs, and/or future needs. Just get in touch if this is still an option [email protected] and I'll be sure to take care of you.
Ignore it. It serves no purpose in testing the VPS.
It does serve the purpose though to find out how much the results can be trusted, how quickly the target server reacts. I'm not sure because I rarely know the target servers, but I have experienced situations where a server with a fat pipe (10+ Gb/s) for whatever reason was "sleepy". web ping helps to recognize that kind of problems and similar.
But yes, I basically just leave in the results because some users might be curious and because it can help to have a bit of context.
As "threatened" here's the DB benchmark. To avoid misunderstandings: this is a very simple benchmark that simply writes about 1 million (actually slightly more) simple records to a sqlite (3) DB. To as far as possible avoid caching and other "trickery" all record data consist of purely random data (which are created upfront and not timed). More details (e.g. the record structure) can be found in my MassiveGrid benchmark & review https://lowendtalk.com/discussion/198672/a-quick-review-benchmark-of-massivegrid-promo-vps.
Here's a (slightly edited) excerpt to provide some context:
RackNerd VPS (2 GB mem., in Dublin):
I excluded the c1vhosting VPS because its performance was ridiculously slow anyway.
Conclusion: The RackNerd VPS benchmarked in this thread is significantly faster then the best of the other VPSs tested prior.
But again, the RN doesn't get the winner badge alone because while it performed best those results are based on an NVMe drive while the host_c VPS achieved its results on a storage drive, i.e. spinning rust (but excellently configured).
TL;DR So, no gold medal but 2 silver medals for both the RN VPS and the host_c VPS.
One thing is clear though - and confirmed again: the RN has a bloody fast drive.
P.S. If you want me to do the DB test with some other VPS (obviously only ones I have) or with different parameters (e.g. larger record size, fewer or more records, different "batch sizes") let me know and I'll do it if possible and sensible.
Btw, can someone tell me how to put some text like e.g. "there" in a link with the Vanilla software?
You're saying you're profiling the server with a single packet response time? You're testing a buffer of a speedtest server that may or may not be conducting simultaneous bidirectional tests on 10Gbps links.
Sign out
Nope, not with a single packet response time. And I'm not "profiling" a server but one specific aspect of a server (being contacted).
the result is so low, are they limit the core to 20%?
result of my vps
cpubenchmark single thread rating