All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
seriesn's new EntryBytes BOX benchmark.
As @seriesn doesn't just blabla about transparency but actually lives it I had an opportunity to test one of the EntryBytes VPSs. Thank you seriesn for providing it!
The results are based on a bit over 100 runs in the NL location.
Machine: amd64, Arch.: amd64, Model: Intel(R) Xeon(R) CPU E5-2620 v4 @ 2.10GHz
OS, version: FreeBSD 12.2, Mem.: 3.986 GB
CPU - Cores: 2, Family/Model/Stepping: 6/79/1
Cache: 32K/32K L1d/L1i, 256K L2, 20M 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
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
----- Processor and Memory -----
ProcMem SC [MB/s]: avg 256.6 - min 218.2 (85.0 %), max 269.0 (104.8 %)
ProcMem MC [MB/s]: avg 524.0 - min 445.4 (85.0 %), max 542.3 (103.5 %)
--- Disk - Buffered ---
Write seq. [MB/s]: avg 1026.30 - min 923.67 (90.0%), max 1145.51 (111.6%)
Write rnd. [MB/s]: avg 4893.79 - min 3822.89 (78.1%), max 6521.84 (133.3%)
Read seq. [MB/s]: avg 2597.72 - min 2434.42 (93.7%), max 3237.45 (124.6%)
Read rnd. [MB/s]: avg 4341.32 - min 4010.14 (92.4%), max 4851.99 (111.8%)
--- Disk - Sync/Direct ---
Write seq. [MB/s]: avg 56.09 - min 31.98 (57.0%), max 84.03 (149.8%)
Write rnd. [MB/s]: avg 142.21 - min 64.73 (45.5%), max 193.04 (135.7%)
Read seq. [MB/s]: avg 1964.65 - min 1691.76 (86.1%), max 2147.82 (109.3%)
Read rnd. [MB/s]: avg 234.02 - min 197.53 (84.4%), max 291.63 (124.6%)
--- Network ---
US LAX lax.download.datapacket.com
DL [Mb/s]: avg 44.15 - min 0.00 (0.0%), max 47.99 (108.7%)
Ping [ms]: avg 144.7 - min 0.0 (0.0%), max 150.2 (103.8%)
Web ping [ms]: avg 150.9 - min 0.0 (0.0%), max 789.1 (523.0%)
NO OSL speedtest.osl01.softlayer.com
DL [Mb/s]: avg 203.52 - min 0.00 (0.0%), max 217.44 (106.8%)
Ping [ms]: avg 24.5 - min 0.0 (0.0%), max 31.9 (130.3%)
Web ping [ms]: avg 58.0 - min 0.0 (0.0%), max 1264.2 (2178.1%)
US SJC speedtest.sjc01.softlayer.com
DL [Mb/s]: avg 35.31 - min 0.00 (0.0%), max 43.60 (123.5%)
Ping [ms]: avg 147.6 - min 145.2 (98.4%), max 150.5 (102.0%)
Web ping [ms]: avg 147.9 - min 145.2 (98.2%), max 150.6 (101.8%)
AU MEL speedtest.c1.mel1.dediserve.com
DL [Mb/s]: avg 24.62 - min 0.00 (0.0%), max 26.73 (108.6%)
Ping [ms]: avg 264.7 - min 0.0 (0.0%), max 269.5 (101.8%)
Web ping [ms]: avg 264.8 - min 0.0 (0.0%), max 272.3 (102.8%)
JP TOK speedtest.tokyo2.linode.com
DL [Mb/s]: avg 24.96 - min 0.00 (0.0%), max 27.58 (110.5%)
Ping [ms]: avg 240.3 - min 240.1 (99.9%), max 241.2 (100.4%)
Web ping [ms]: avg 255.4 - min 240.2 (94.0%), max 1211.0 (474.1%)
IT MIL speedtest.mil01.softlayer.com
DL [Mb/s]: avg 222.84 - min 211.44 (94.9%), max 241.99 (108.6%)
Ping [ms]: avg 23.1 - min 23.0 (99.4%), max 23.9 (103.3%)
Web ping [ms]: avg 66.9 - min 23.0 (34.4%), max 1471.0 (2197.9%)
FR PAR speedtest.par01.softlayer.com
DL [Mb/s]: avg 445.23 - min 369.58 (83.0%), max 468.39 (105.2%)
Ping [ms]: avg 12.8 - min 12.6 (98.3%), max 14.5 (113.1%)
Web ping [ms]: avg 52.4 - min 12.7 (24.2%), max 1355.5 (2585.6%)
SG SGP mirror.sg.leaseweb.net
DL [Mb/s]: avg 36.76 - min 0.00 (0.0%), max 41.26 (112.2%)
Ping [ms]: avg 158.6 - min 158.3 (99.8%), max 160.6 (101.2%)
Web ping [ms]: avg 159.5 - min 158.3 (99.2%), max 237.7 (149.0%)
BR SAO speedtest.sao01.softlayer.com
DL [Mb/s]: avg 23.89 - min 0.00 (0.0%), max 31.93 (133.6%)
Ping [ms]: avg 187.5 - min 184.9 (98.6%), max 190.7 (101.7%)
Web ping [ms]: avg 211.0 - min 184.9 (87.6%), max 1430.1 (677.7%)
IN CHN speedtest.che01.softlayer.com
DL [Mb/s]: avg 30.47 - min 0.00 (0.0%), max 36.67 (120.3%)
Ping [ms]: avg 167.6 - min 0.0 (0.0%), max 188.1 (112.2%)
Web ping [ms]: avg 189.1 - min 0.0 (0.0%), max 994.8 (526.1%)
GR UNK speedtest.ftp.otenet.gr
DL [Mb/s]: avg 83.43 - min 0.00 (0.0%), max 144.71 (173.4%)
Ping [ms]: avg 24.5 - min 0.0 (0.0%), max 53.4 (217.5%)
Web ping [ms]: avg 30.8 - min 0.0 (0.0%), max 616.1 (2001.6%)
US WDC mirror.wdc1.us.leaseweb.net
DL [Mb/s]: avg 66.13 - min 0.00 (0.0%), max 73.08 (110.5%)
Ping [ms]: avg 92.5 - min 0.0 (0.0%), max 97.1 (105.0%)
Web ping [ms]: avg 92.7 - min 0.0 (0.0%), max 97.1 (104.7%)
RU MOS speedtest.hostkey.ru
DL [Mb/s]: avg 114.94 - min 0.00 (0.0%), max 136.74 (119.0%)
Ping [ms]: avg 47.2 - min 0.0 (0.0%), max 52.5 (111.2%)
Web ping [ms]: avg 48.7 - min 0.0 (0.0%), max 69.4 (142.5%)
US DAL speedtest.dal05.softlayer.com
DL [Mb/s]: avg 44.09 - min 0.00 (0.0%), max 54.41 (123.4%)
Ping [ms]: avg 116.7 - min 116.3 (99.7%), max 118.9 (101.9%)
Web ping [ms]: avg 117.0 - min 116.3 (99.4%), max 135.6 (115.9%)
UK LON speedtest.lon02.softlayer.com
DL [Mb/s]: avg 574.97 - min 381.26 (66.3%), max 699.18 (121.6%)
Ping [ms]: avg 8.5 - min 8.4 (98.3%), max 8.8 (103.0%)
Web ping [ms]: avg 37.1 - min 8.5 (22.9%), max 1246.0 (3362.3%)
US NYC nyc.download.datapacket.com
DL [Mb/s]: avg 72.46 - min 0.00 (0.0%), max 81.06 (111.9%)
Ping [ms]: avg 84.6 - min 0.0 (0.0%), max 89.0 (105.2%)
Web ping [ms]: avg 88.0 - min 0.0 (0.0%), max 126.9 (144.2%)
RO BUC 185.183.99.8
DL [Mb/s]: avg 142.72 - min 0.00 (0.0%), max 181.58 (127.2%)
Ping [ms]: avg 33.4 - min 0.0 (0.0%), max 38.8 (116.1%)
Web ping [ms]: avg 44.5 - min 0.0 (0.0%), max 392.3 (881.0%)
CN_HK mirror.hk.leaseweb.net
DL [Mb/s]: avg 30.41 - min 0.00 (0.0%), max 35.65 (117.2%)
Ping [ms]: avg 191.3 - min 190.7 (99.7%), max 191.4 (100.1%)
Web ping [ms]: avg 194.3 - min 190.7 (98.2%), max 207.5 (106.8%)
DE FRA fra.lg.core-backbone.com
DL [Mb/s]: avg 607.65 - min 524.45 (86.3%), max 698.84 (115.0%)
Ping [ms]: avg 7.7 - min 7.6 (98.6%), max 9.8 (127.1%)
Web ping [ms]: avg 7.7 - min 7.6 (98.5%), max 9.8 (127.0%)
First a meta-remark: In the network tests '0.0' means that some of the test run failed. That may or may not be the providers/products fault. Based on my observations it's usually either the network backends fault or, although rarely, the targets fault. The position of the 'avg' value on the 'min' - 'max' axis usually is a good indicator whether a failure happened only exceptionally or more or less frequently.
First processor & memory. These days one might be tempted to compare everything to a Ryzen 3000 or 5000. I'd like to correct that view which IMO is mistaken; I'm sometimes working with a 2450L dedi and have a 26xx v2 in my lab, and trust me, those machines are definitely not snails. Unless you know very well what you are doing and are doing it very often (having got a "feeling" for a box) -and- push the box really hard you won't see a lot of difference.
That's why I really like the 26xx v4. It's fast. Fast as in about 60% - 90% faster than a 24xx or a 26xx v2 - yet it's much cheaper to buy those machines than to buy Ryzen systems (which of course translates to lower VPS prices). When I had a talk with a provider who offers both, Ryzen (and Epyc) as well as 26xx v4 he confirmed what I had developed as an educated guess: the Ryzen systems are much, much more expensive.
Also a well chosen 26xx v4 isn't far behind an Epyc2. Or in other words: You get almost the performance of an Epyc but at much lower cost. So, seriesn made a smart choice for his EntryBytes systems. And (I almost wrote "of course") you have/get all the attractive flags, like AES, AVX2, popcount and even hypervisor from EntryBytes. Nice. I like that choice.
Next, the disk.I'll keep it short. Not at all bad but neither anything to write home about, but then this is a budget VPS. You get a solid SSD (Raided). If you want really fassst disks you'll need to go one level up to NexusBytes, at a cost difference of course.
Now to the network.
The bad news is that most targets shows failures. The good news is that most of those have only occasional hiccups as the 'avg' value clearly shows (it's usually quite close to 'max'). And I noticed that the network backend seems to be a bit optimized towards preferring the top locations. DE_FRA is a good example with about 600 Mb/s download speed. I personally prefer a more balanced approach but all in all I wouldn't complain, after all we are talking about products within the LET price range. That said, even the "exotic" target offer reasonable speed (which can't be said about every provider ...)
Summary/verdict: Well, I guess nobody expected something sh_tty from seriesn. Of bloody course even their budget line is decent. Whether you find it also attractive ("wanna buy!") probably will depend on mainly 2 factors, (a) price (which I don't know but assume is between nice and reasonable), and (b) the factors a benchmark can't capture; support is a great example, NexusBytes (and certainly EntryBytes too) offers almost incredible support.
Comments
please run yabs - https://github.com/masonr/yet-another-bench-script
and post results here. thx.
You can (or eventually) find those in their offer thread itself.
If I considered yabs to be a benchmark meeting my needs and expectations I hardly would have invested the (quite considerable) efforts to write the vpsbench software, which btw. quite a few providers use internally or ask me to do it for them in order to test their products and/or spot weak spots.
Thanks for noticing that I responded very politely.
General remark (as it occasionally happens): Do not ask me to run [some canned "benchmark"] xyz! Don't! (unless you want me to think "what a clueless idiot!").
lol
Where are the 64-bit network speed measurements?
Oh well, @yoursunny and @Lee ... one is an obvious idiot and the other one behaves like one ...
So, I make efforts to provide a service to the community and some idiots just have to take a dump? Sad. Is that really how we want our community to be?
No sure which one I am, but on LET, I will accept either, probably both
@JSG thank you once again for your generous effort in helping us improving our day to day business. Valid points regarding disk is noted and will be considered during our next build
Super detailed feedbacks and always helpful. Many thanks
Ah yes, nothing better than a whole new level of "btw. I use Arch" coming from FreeBSD user.
how many people understand this benchmark?
YABS pls
It's @jsg, you're not supposed to understand his posts, just be impressed that he can write so much detail.
Francisco
You are warmly welcome.
?? I'm a linux user, sometimes Arch but mostly debian (w/o systemd) plus I'm an Alpine fan. As for FreeBSD, as multiple times explained, I use that for my benchmarks simply because linux is caching (on multiple levels) so aggressively that the disk benchmark numbers are pretty much worthless.
I'm generally quite open wrt OS as I should be as a developer, because the OS(s) is/are not my choice but the client's. I confess though that I try to avoid Windows whenever possible but sometimes a client specs it.
That was a cheap shot and not a particularly smart one at that. I've explained again and again and openly what vpsbench measures, why it measures it and how it measures it.
Yes, my benchmark is a bit more complex than most "quick scripts" but then, computers are complex too and so are customer needs. Looking for a VPS for largely static file serving is quite different from looking for a DB server - and unlike other benchmarks mine does offer the information needed.
And then there are some VPS/VDS specific issues like for example the fact that one (1) test run is pretty much worthless; it's like driving once at some arbitrary time through a city and then making a statement about the traffic situation there. It just doesn't make sense.
Btw. if someone doesn't understand something (s)he can ask; if asked friendly I answer friendly.
But yes, I'm guilty of neither using nor communicating à la twitter. For those who like it super short (albeit shallow) there's plenty of alternatives out there.
What's a cheap shot...? You write novels, you're known for it. It's impressive that you can go into such detail.
Not everyones out to get you boss.
Francisco
If you were looking for some VPSs for a company whose success depends to some degree on the systems you run your stuff on, would you prefer to make your choice based on a one off quick benchmark script or based on detailed and rich data gathered in 100 plus runs at different times of day (and night) and on correct and precise timing? I'd hope the latter ...
Have you considered open-sourcing your script?
Didn't claim your tests were poorly done or that they shouldn't have been done. I said you go into huge detail.
Some people want a YABS. Some people want the Art of War.
Francisco
no YABS no PARTY
It might not be an equivalent to YABS, but I see the script's output is fairly self-explanatory.
...would you do so on LOWENDtalk?
please explain to me, my dear watson
Actually I had v 1.x open sourced. I stopped that with v 2.x because (not really surprisingly) the "we want the source!" demands were but herd signalling and almost nobody downloaded the source code.
Btw, vpsbench is not a script but a real program written in a decent statically typed (compiled) language.
Thank you. I think so too and have in fact invested quite a bit of work into presenting the data in a decent way. But I have to confess that I'm quite bad at presenting things nicely (and web design, and ...).
(a) granted the example was chosen for contrast. But I do know people who want to chose/select wisely even for private stuff.
(b) Yes, quite a few small companies do and so do quite some companies in poorer countries.
Sorry I cannot understand how it indicates the frequency of failure.
Does the "avg" includes the failure which is counted as 0ms, so that with high chance of failure, "avg" becomes smaller?
Could you please show the failure percentage explicitly?
Well, as luck has it, I'm already working on that