All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
AVSISP (Macedonia) VPS benchmark and review
While I'm usually not interested in VPS for >= €5/mo this one got me interested for two reasons, (a) the "all HW doubled", and (b) the kind of exotic location, Macedonia. So I clicked "buy".
A third factor would have been the unlimited traffic at 1 Gb/s if I were interested in that; I'm not, but in case you are the price tag would become more digestable (as in "if I needed a VPS with high traffic I'd consider €5/mo acceptable, maybe even attractive").
What you see here is based on a bit over 30 runs over more than 24 hrs.
First sysinfo, processor and memory
Version 2.5.0a, (c) 2018+ jsg (->lowendtalk.com)
Machine: amd64, Arch.: amd64, Model: AMD EPYC 7302 16-Core Processor
OS, version: FreeBSD 14.4, Mem.: 1.989 GB
CPU - Cores: 2, Family/Model/Stepping: 23/49/0
Cache: 64K/64K L1d/L1i, 512K 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 htt sse3 pclmulqdq ssse3 fma cx16
sse4_1 sse4_2 popcnt aes xsave osxsave avx f16c rdrnd hypervisor
Ext. Flags: syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm lahf_lm cmp_legacy svm
cr8_legacy lzcnt sse4a misalignsse 3dnowprefetch osvw perfctr_core
AES? Yes
Nested Virt.? Yes
HW RNG? Yes
ProcMem SC [MB/s]: avg 120.3 - min 51.6 (42.9 %), max 212.7 (176.9 %)
ProcMem MA [MB/s]: avg 463.2 - min 353.0 (76.2 %), max 561.9 (121.3 %)
ProcMem MB [MB/s]: avg 555.3 - min 423.4 (76.3 %), max 695.1 (125.2 %)
ProcMem AES [MB/s]: avg 717.6 - min 615.4 (85.8 %), max 986.2 (137.4 %)
ProcMem RSA [kp/s]: avg 55.9 - min 51.6 (92.3 %), max 63.9 (114.3 %)
IMO a bit meh for the price. That's Xeon E5 v3 territory not Epyc, plus it doesn't have AVX2 and the crypto results are Xeon E5v4 like.
I'd also like to add that during the first 10 or so runs performance was way better (e.g. SC 180 to 190) but later obviously some other users on the node yabs'd the whole node down.
Maybe the 40GB disk (NVMe) makes up for it?
--- Disk 4 KB - Buffered ---
Write seq. [MB/s]: avg 6.72 - min 5.88 (87.6%), max 7.33 (109.2%)
Write rnd. [MB/s]: avg 6.65 - min 6.00 (90.3%), max 7.28 (109.5%)
Read seq. [MB/s]: avg 21.16 - min 19.97 (94.4%), max 22.85 (108.0%)
Read rnd. [MB/s]: avg 18.76 - min 16.97 (90.4%), max 20.87 (111.2%)
--- Disk 4 KB - Sync/Direct ---
Write seq. [MB/s]: avg 6.75 - min 6.21 (92.0%), max 7.37 (109.2%)
Write rnd. [MB/s]: avg 6.70 - min 6.31 (94.2%), max 7.09 (105.8%)
Read seq. [MB/s]: avg 21.03 - min 19.89 (94.6%), max 22.50 (107.0%)
Read rnd. [MB/s]: avg 18.92 - min 17.81 (94.1%), max 21.04 (111.2%)
--- Disk 64 KB - Buffered ---
Write seq. [MB/s]: avg 58.83 - min 54.34 (92.4%), max 63.49 (107.9%)
Write rnd. [MB/s]: avg 71.84 - min 66.77 (92.9%), max 79.78 (111.1%)
Read seq. [MB/s]: avg 1853.96 - min 1669.49 (90.0%), max 2119.36 (114.3%)
Read rnd. [MB/s]: avg 241.54 - min 214.44 (88.8%), max 262.53 (108.7%)
--- Disk 64 KB - Sync/Direct ---
Write seq. [MB/s]: avg 5.02 - min 4.42 (88.0%), max 6.06 (120.7%)
Write rnd. [MB/s]: avg 2.56 - min 2.35 (91.8%), max 3.30 (129.0%)
Read seq. [MB/s]: avg 1842.14 - min 1591.06 (86.4%), max 2120.76 (115.1%)
Read rnd. [MB/s]: avg 237.44 - min 201.85 (85.0%), max 265.57 (111.8%)
--- Disk 1 MB - Buffered ---
Write seq. [MB/s]: avg 84.36 - min 74.29 (88.1%), max 93.70 (111.1%)
Write rnd. [MB/s]: avg 185.03 - min 157.21 (85.0%), max 204.04 (110.3%)
Read seq. [MB/s]: avg 2558.12 - min 2200.78 (86.0%), max 3138.35 (122.7%)
Read rnd. [MB/s]: avg 930.33 - min 754.23 (81.1%), max 1075.04 (115.6%)
--- Disk 1 MB - Sync/Direct ---
Write seq. [MB/s]: avg 23.32 - min 19.75 (84.7%), max 27.27 (117.0%)
Write rnd. [MB/s]: avg 21.79 - min 20.15 (92.5%), max 24.89 (114.2%)
Read seq. [MB/s]: avg 2417.58 - min 2118.71 (87.6%), max 2802.89 (115.9%)
Read rnd. [MB/s]: avg 919.49 - min 823.15 (89.5%), max 1056.14 (114.9%)
--- Disk IOps (Sync/Direct) ---
Write seq. [MB/s]: avg 45.77 - min 37.78 (82.5%), max 49.93 (109.1%)
IOps : avg 11717.77 - min 9671.49 (82.5%), max 12781.80 (109.1%)
Yes, it does and how well it does! a bit over 45 MB/s and well over 10k IOps plus a relatively low spread is an excellent result.
Macedonia doesn't seem to promise good connectivity but I'm certainly ready for a nice surprise. Let's see ...
--- Europe ---
NO OSL mirror.terrahost.no [F: 0]
DL [Mb/s]: avg 210.6 - min 81.8 (38.9%), max 228.7 (108.6%)
Ping [ms]: avg 51.4 - min 51.2 (99.6%), max 52.2 (101.6%)
Web ping [ms]: avg 113.8 - min 51.2 (45.0%), max 197.3 (173.3%)
UK LON lon.speedtest.clouvider.net [F: 0]
DL [Mb/s]: avg 242.1 - min 165.9 (68.5%), max 250.8 (103.6%)
Ping [ms]: avg 44.9 - min 44.8 (99.8%), max 45.5 (101.4%)
Web ping [ms]: avg 45.6 - min 44.8 (98.3%), max 48.5 (106.4%)
NL AMS nl.mirrors.clouvider.net [F: 0]
DL [Mb/s]: avg 281.5 - min 177.7 (63.1%), max 306.8 (109.0%)
Ping [ms]: avg 36.8 - min 36.6 (99.4%), max 37.0 (100.5%)
Web ping [ms]: avg 43.7 - min 36.7 (83.9%), max 82.6 (188.9%)
DE FRA fra.lg.core-backbone.com [F: 0]
DL [Mb/s]: avg 357.0 - min 331.0 (92.7%), max 372.9 (104.5%)
Ping [ms]: avg 32.3 - min 32.2 (99.7%), max 32.4 (100.3%)
Web ping [ms]: avg 32.4 - min 32.2 (99.5%), max 33.6 (103.8%)
FR PAR mirror.in2p3.fr [F: 0]
DL [Mb/s]: avg 279.9 - min 278.5 (99.5%), max 281.6 (100.6%)
Ping [ms]: avg 39.4 - min 39.4 (99.9%), max 39.5 (100.2%)
Web ping [ms]: avg 62.2 - min 39.8 (64.0%), max 98.8 (159.0%)
CH ZUR mirror.metanet.ch [F: 0]
DL [Mb/s]: avg 271.4 - min 136.7 (50.4%), max 306.3 (112.8%)
Ping [ms]: avg 38.0 - min 37.5 (98.7%), max 40.8 (107.4%)
Web ping [ms]: avg 55.5 - min 37.5 (67.6%), max 184.7 (333.0%)
ES MAD mirror.raiolanetworks.com [F: 0]
DL [Mb/s]: avg 217.0 - min 212.8 (98.1%), max 220.0 (101.4%)
Ping [ms]: avg 51.2 - min 51.1 (99.8%), max 51.3 (100.2%)
Web ping [ms]: avg 53.2 - min 52.1 (97.9%), max 54.5 (102.4%)
RO BUC mirrors.hosterion.ro [F: 0]
DL [Mb/s]: avg 786.8 - min 747.9 (95.1%), max 827.2 (105.1%)
Ping [ms]: avg 12.1 - min 12.0 (99.2%), max 12.2 (100.8%)
Web ping [ms]: avg 12.4 - min 12.2 (98.3%), max 13.2 (106.3%)
RU MOS speedtest.hostkey.ru [F: 0]
DL [Mb/s]: avg 169.2 - min 163.4 (96.6%), max 172.9 (102.2%)
Ping [ms]: avg 65.1 - min 65.0 (99.8%), max 65.3 (100.3%)
Web ping [ms]: avg 80.0 - min 66.1 (82.6%), max 85.6 (106.9%)
--- Asia / Oceania ---
RU SIB mirror.truenetwork.ru [F: 0]
DL [Mb/s]: avg 69.2 - min 21.3 (30.8%), max 109.9 (158.8%)
Ping [ms]: avg 104.1 - min 102.3 (98.3%), max 105.0 (100.9%)
Web ping [ms]: avg 113.6 - min 102.5 (90.2%), max 143.5 (126.3%)
IN MUM mirrors.piconets.webwerks.in [F: 0]
DL [Mb/s]: avg 83.2 - min 77.8 (93.4%), max 85.9 (103.2%)
Ping [ms]: avg 140.6 - min 139.2 (99.0%), max 181.7 (129.2%)
Web ping [ms]: avg 150.6 - min 139.2 (92.4%), max 251.4 (166.9%)
SG SGP mirror.jingk.ai [F: 0]
DL [Mb/s]: avg 64.6 - min 62.5 (96.6%), max 65.4 (101.2%)
Ping [ms]: avg 181.0 - min 180.7 (99.8%), max 182.4 (100.8%)
Web ping [ms]: avg 267.9 - min 189.9 (70.9%), max 533.5 (199.2%)
CN HKG mirrors.xtom.hk - Failed
CN BEJ mirrors.bfsu.edu.cn [F: 1]
DL [Mb/s]: avg 38.3 - min 0.0 (0.0%), max 45.5 (118.7%)
Ping [ms]: avg 307.7 - min 0.0 (0.0%), max 351.8 (114.3%)
Web ping [ms]: avg 307.7 - min 0.0 (0.0%), max 351.8 (114.3%)
JP TOK ftp.udx.icscoe.jp [F: 0]
DL [Mb/s]: avg 40.7 - min 36.1 (88.7%), max 42.9 (105.4%)
Ping [ms]: avg 269.5 - min 269.4 (99.9%), max 272.3 (101.0%)
Web ping [ms]: avg 272.4 - min 269.4 (98.9%), max 280.3 (102.9%)
AU SYD mirrors.xtom.au [F: 0]
DL [Mb/s]: avg 43.0 - min 42.3 (98.5%), max 43.7 (101.7%)
Ping [ms]: avg 262.7 - min 262.7 (100.0%), max 262.8 (100.0%)
Web ping [ms]: avg 328.2 - min 313.5 (95.5%), max 404.1 (123.1%)
--- Africa ---
KE NAI mirror.liquidtelecom.com [F: 1]
DL [Mb/s]: avg 41.8 - min 0.0 (0.0%), max 46.4 (111.0%)
Ping [ms]: avg 246.8 - min 242.7 (98.3%), max 251.1 (101.7%)
Web ping [ms]: avg 253.2 - min 242.7 (95.8%), max 284.2 (112.2%)
ZA, WEC archlinux.za.mirror.allworldit.com [F: 4]
DL [Mb/s]: avg 44.7 - min 0.0 (0.0%), max 59.3 (132.7%)
Ping [ms]: avg 215.0 - min 0.0 (0.0%), max 346.4 (161.1%)
Web ping [ms]: avg 225.0 - min 0.0 (0.0%), max 346.4 (154.0%)
--- Americas ---
CA MTL speedtest.mtl2.ca.leaseweb.net [F: 0]
DL [Mb/s]: avg 86.1 - min 29.3 (34.0%), max 96.3 (111.9%)
Ping [ms]: avg 117.4 - min 117.0 (99.6%), max 118.0 (100.5%)
Web ping [ms]: avg 118.6 - min 117.1 (98.7%), max 120.7 (101.7%)
US NYC nyc.mirrors.clouvider.net [F: 0]
DL [Mb/s]: avg 97.9 - min 94.4 (96.4%), max 99.5 (101.6%)
Ping [ms]: avg 116.6 - min 116.5 (99.9%), max 116.8 (100.2%)
Web ping [ms]: avg 119.7 - min 116.5 (97.3%), max 156.6 (130.8%)
US ASH ash.speedtest.clouvider.net [F: 1]
DL [Mb/s]: avg 82.9 - min 0.0 (0.0%), max 92.0 (111.0%)
Ping [ms]: avg 133.9 - min 127.9 (95.5%), max 184.0 (137.4%)
Web ping [ms]: avg 140.5 - min 127.9 (91.1%), max 315.5 (224.6%)
US PIB mirror.pit.teraswitch.com [F: 0]
DL [Mb/s]: avg 79.2 - min 35.8 (45.2%), max 90.9 (114.8%)
Ping [ms]: avg 122.3 - min 0.0 (0.0%), max 127.0 (103.9%)
Web ping [ms]: avg 127.7 - min 0.0 (0.0%), max 220.1 (172.3%)
US MIA speedtest.mia11.us.leaseweb.net [F: 5]
DL [Mb/s]: avg 62.9 - min 0.0 (0.0%), max 80.3 (127.6%)
Ping [ms]: avg 151.5 - min 147.5 (97.4%), max 165.8 (109.5%)
Web ping [ms]: avg 151.5 - min 147.5 (97.4%), max 165.8 (109.4%)
US CHI ord.mirror.rackspace.com [F: 0]
DL [Mb/s]: avg 85.9 - min 83.4 (97.1%), max 89.2 (103.8%)
Ping [ms]: avg 129.6 - min 129.4 (99.8%), max 134.1 (103.4%)
Web ping [ms]: avg 151.2 - min 130.9 (86.6%), max 205.8 (136.1%)
US ATL atl.speedtest.clouvider.net [F: 0]
DL [Mb/s]: avg 83.6 - min 81.7 (97.8%), max 86.0 (102.8%)
Ping [ms]: avg 136.3 - min 136.2 (99.9%), max 136.4 (100.1%)
Web ping [ms]: avg 138.0 - min 136.2 (98.7%), max 149.4 (108.3%)
US PHO speedtest.phx1.us.leaseweb.net [F: 0]
DL [Mb/s]: avg 66.3 - min 64.5 (97.4%), max 68.1 (102.8%)
Ping [ms]: avg 174.9 - min 174.6 (99.8%), max 175.8 (100.5%)
Web ping [ms]: avg 175.0 - min 174.6 (99.8%), max 175.9 (100.5%)
US PTL mirrors.cat.pdx.edu [F: 0]
DL [Mb/s]: avg 61.3 - min 30.2 (49.3%), max 63.5 (103.5%)
Ping [ms]: avg 184.0 - min 183.4 (99.7%), max 197.1 (107.1%)
Web ping [ms]: avg 185.8 - min 183.4 (98.7%), max 197.1 (106.1%)
US LAX la.speedtest.clouvider.net [F: 0]
DL [Mb/s]: avg 61.2 - min 59.4 (97.1%), max 64.0 (104.6%)
Ping [ms]: avg 183.4 - min 183.3 (99.9%), max 184.3 (100.5%)
Web ping [ms]: avg 186.9 - min 183.3 (98.1%), max 201.4 (107.8%)
US SJO mirrors.xtom.us [F: 0]
DL [Mb/s]: avg 61.9 - min 60.0 (96.8%), max 63.9 (103.2%)
Ping [ms]: avg 179.2 - min 179.2 (100.0%), max 179.3 (100.0%)
Web ping [ms]: avg 181.7 - min 179.2 (98.6%), max 193.8 (106.7%)
US SEA speedtest.sea11.us.leaseweb.net [F: 4]
DL [Mb/s]: avg 54.0 - min 0.0 (0.0%), max 64.5 (119.5%)
Ping [ms]: avg 180.8 - min 180.7 (99.9%), max 181.1 (100.2%)
Web ping [ms]: avg 180.8 - min 180.7 (99.9%), max 181.1 (100.2%)
BR SPA mirrors.ic.unicamp.br [F: 3]
DL [Mb/s]: avg 37.1 - min 0.0 (0.0%), max 44.7 (120.2%)
Ping [ms]: avg 259.3 - min 254.2 (98.0%), max 275.0 (106.0%)
Web ping [ms]: avg 261.8 - min 254.7 (97.3%), max 276.0 (105.4%)
CL SAN elmirror.cl [F: 0]
DL [Mb/s]: avg 45.9 - min 36.0 (78.6%), max 48.0 (104.6%)
Ping [ms]: avg 244.9 - min 242.0 (98.8%), max 273.7 (111.7%)
Web ping [ms]: avg 278.9 - min 242.2 (86.8%), max 516.0 (185.0%)
Front-up: Almost all secondary targets were reached. That's a good omen.
Europe - Pretty much all of Europe over 200 Mb/s, DE Fra even about 350 Mb/s, and RO Buc over 750 Mb/s.
Sadly all routing goes via derpent (cogent) for at least 2 hops (Skopje and Sofia) but all to the east of France then goes mainly via twelve99, except for France, Italy, and Spain to which packets have to endure a longer derpent torture (until Milano or Marseille). "Funnily" Romania targets go via derpent all the way, and in one case - typical derpent idiocy - via a DE Fra detour, yuck. The russian targets are mistreated even worse.
Asia/Oceania - India is routed via Marseille (and then via Tata or some other undersea fiber), Singapore is interesting because one target (jingk.ai) is routed all the way via derpent, certainly not great but neither horribly bad) while the other SGP target (sg.gs) is utterly fucked up by derpent using the us-american idiot route, adding over 125 ms, yuck! Can someone please clean out all the drugs from derpent's "NOC"?!
The JP route gets the same imbecile treatment, but the routes to China are way better because some other - and real - carrier, chinese presumably, grabs the packets in DE Fra and transports them to China mainland within about 100 - 120 ms (for a total of about 250 - 300 ms). Similarly with Ozzyland; there a real carrier (GSL) grabs the packets early in BG Sofia.
All in all the results for Asia/Oceania are mostly mediocre, thanks to fucking derpent.
Africa - mediocre as well
Americas - Better but still not at all good. A positive side note: Towards @Clouvider, xTom, and Rackspace targets the packets tend to get grabbed early in BG Sofia by twelve99. Pretty much all other targets, sadly incl. Leaseweb, go at least to the other side of the ocean or even all the way via derpent.
Sadly, the results for LATAM are mediocre as well; while all targets are reached derpent make sure that they are routed badly and slowly (via Miami).
Summary/verdict: a boring not at all fast processor and memory, a remarkably great disk, and mediocre connectivity. To be fair, I guess in Macedonia there aren't too many options in terms of international carriers.
All in all I will not keep this VPS but won't say "stay away" because unlimited traffic albeit with mediocre connectivity for €5/mo may be a relatively good deal for some.
In ServerVerify lingo I give it 3 points/stars out of five.
Comments
Glad you finally decided to try it yourself.
I'm not quite sure I should even respond to this. It seems like the server itself is great, the processor isn't the latest one off the press (because we'd have to charge a lot more.than 5 - due to RAM prices which have to be offset somewhere) but is actually way better than an E5 as you claim (benchmarks prove double the E5 standard results except E5-1660v4 which does above it) and you REALLY don't like Cogent - which is your opinion I guess - however they are the largest T1 on the planet in terms of reach on-net.... And you're correct - there aren't many options in Macedonia... We're looking at adding others, but in Macedonia, like much of the non-Bulgaria Balkans, you're left with "get what you can, there arent options" usually.
Overall - I'm sorry to say that this is an extremely biased review that focuses on network being Cogent (which is kind of obvious, check our ASN - Cogent is only upstream on Macedonia IPs). And you're comparing I would assume extremely large providers like Amazon that can afford the latest and greatest processors and to lose money on smaller VMs to sell to huge governments and all at a massive markup to a normal VPS provider in a niche market - which isn't at all a fair comparison (about CPU especially).
New popcorn 🍿
I agree to some degree, for instance re the current processor and memory situation and evidently the very limited carrier options in the region.
However, I disagree with your "biased" accusation. For one my reviews always are based on the data I gathered. Plus, in your case, no offense intended, I didn't even know you before I purchased that VPS.
Yes, I very much dislike derpent (cogent) - but for a reason: whenever I see derpent's routing I see crap, incl. in particular the imbecile adding of more than 100 ms and quite a few hops by absolutely needlessly, and in fact counter-productive, routing via North-America.
Now, again, I do see - and in fact assumed and wrote - that your carrier choice is quite limited, and clearly you are not guilty for what derpent does.
But you see, at the end of the day, your users usually just see the outcome, and so do I.
What I saw was an Epyc with a performance like a Xeon E5 v3 or v4, mediocre connectivity, and an excellent disk - and that's what I wrote.
Yes, the current market situation is difficult but that doesn't change the facts and hard data.
Let me assure you that I'm not at all biased against you. The major problem IMO is the price because this LET and bang per buck is important here.
Yes, the but you also have to realize you're running that benchmark at the same time as ~30 users who purchased their VMs over the past day or 2 and were also running benchmarks.
The processor is actually quite good. But with many tests running at once, of course you'll see some decreases. If you look at other hosts in the same region, you're going to find that the price of 5/mo is EXTEMELY low. Most are selling at 25/mo a VM - smallest package.
We keep our prices the same over several regions. Which means in some regions, we may seem expensive (NL node for example seems expensive for that region because there's a lot of competitors that can offer lower prices). But for Macedonia, I can assure you we are one of the cheapest on the market. In the bottom 5%.
Back to processors - check the processor model - see the scores that are publicly listed - compare it to an E5 of same Gen. You'll see that this is a MASSIVE improvement over an E5.
By biased, I wasn't saying you as a person are biased or that you even represent a single company which you make look better or something. I was referring to your 80% of review being about Cogent instead of about the server or hardware.
Several users have posted YABS that exceed your results by more than double. This leads me to believe that you ran it while other users also were running it, causing lower results.
For an unbiased review that fully reflects the situation, wait until about 2-3 weeks AFTER a launch or deal, when things are stable and grounded in that location, and then perform your testing.
For example, we plan to upgrade network, add IXPs, etc locally - but we always launch first, then upgrade. Check our history with any other launch. You'll see they launched with a single upstream and ended up in multiple IXPs, routing made excellent through manual route interventions and testing, etc. if your test was done in say 3 weeks, you would have seen a lot different results than "just Cogent" and probably not ranted about Cogent for ~80% of the text of the review.
No disrespect at all meant here - just stating facts.
oops.
Eh, let everyone have a popcorn moment once in a while 🍿
Ive just tested speeds across europe for comparison @avsisp @jsg
Thanks
. Looks like a lot more than 100-250Mb/s. Hmm 🤔
When I say the processor and memory performance is in Xeon E5v3 or v4 territory that is based on plenty other benchmarks I did.
What I agree with is that obviously during my testing other users came onboard and yabs'd the node down - and I did write that.
However the situation is very similar with other VPS and other providers. And I'm absolutely willing to run my benchmark again in say, two weeks. Unfortunately I can not test the VPS again in three or more weeks as I've paid for only one month. If you let me keep the VPS somewhat longer for free I'll gladly wait for you connectivity upgrade(s) and run the benchmarks again; of course I wouldn't use that VPS for anything but said testing.
Nice attempt
I only have a single question...
Is 3 weeks more or less than a month?
OK, after those two comments I take back my offer to you that I made with friendly intentions.
I dont think biased is the right way to describe this review.
The assumption that the node was slow because many users were benchmarking is just an assumption. Theres no concrete evidence presented for that so using it to dismiss the results is not valid.
On the network side saying ther are limited upstream options in the region is understandable from an infrastructure perspective but from a customer point of view it doesnt really change expectations. A paying user isnt evaluating how difficult the region is they are evaluating the performance they actually get.
The same applies to pricing. Bringing up that others charge 25USD does not really address the core feedback. Setting the price at 5USD is your business decision to capture a market gap, which is totally fair, but once you do that users are still going to judge the service on performance.
Else add a point that dont expect good network performance because we dont have good upstream options and we are selling at 5USD.
@avsisp you should really have taken it as a feedback and put your point in a subtle way rather than trying to dismiss it.
This was really rude... It changed my perspective towards your brand.
I understand where you're coming from - however a review is one thing and a post simply ranting about Cogent is another.
The entire "review" as basically a claim of Cogent's incompetence and has no reflection on us as a provider or our services as they stand. Anyone who knows us knows we aren't fans of Cogent ourselves for the same reasons basically stated in a long rant there - long random detours at times.
It's hard to get an ounce of feedback to "work on" or upgrade from when the entire post is just Cogent this, Cogent that.
They could have simply posted this with title "Cogent network rant" and it would have been a lot more relevant and on-topic.
Several users post comments with YABS, discuss performance, compare it to others, offer constructive criticisms, etc which are always welcomed and acted upon.
And finally - in regards to your thinking that was rude - it was not - I for one am a very direct person and it was simply a direct reply to his statement. He said that he would be glad to test again in a few weeks but would need an extension for free because "he only purchased for a month" - 3 weeks is 1 week less than a month. Aka - his statement made no sense at all and broke all logic. The question was intended to get him to see the mistake of his math / logic there and say "oh, sorry - I didn't think about that" or acknowledge that it was incorrect - not to be "rude". Rude would be us calling them names, pointing fingers, using bad words, etc. None of which has happened here. It's been simply replies to their claims. Including 1 entirely contradictory to their report YABS from a LONG TIME user showing even their claimed internet speeds were entirely off-base.
He can't test it again in 3 weeks because his month will be up in 2 weeks, are you joking right now?
You don't want to show your ass here because we will take pictures and remind you of it until the universe ends.
flopped hard with this one ngl
I mean, the Macedonia launch was posted on April 16th. Assuming they purchased it on that day, they should have ~24 more days left - which is well within the three-week mark.
It doesn't matter though - I think this discussion stopped being in good faith and productive long ago.
Pardon my french but that's really BS!
It's certainly no me who took the decision to route packets to several targets in Asia via North-America. I'm just the one who noticed and it spelled it out.
And NO, my review was not a derpent rant, it merely spelled out a very major reason why your connectivity is mediocre.
Edit:
@zed @skimply153 I purchased it a few days ago and quite likely would have enough time left to squeeze another benchmark run in before my VPS expires. But (a) what for? Just to get drawn into another fight with that provider? And (b) why "three weeks"? Does he really know for sure when exactly his update (he talked about) will be completed? Btw. I work for a living and do not always have time right away.
All in all I think that provider didn't do himself a favour. I'm known for being unbiased (hell, I once gave a provider whom I hated a very positive review because his product deserved it) and a smart and decent provider wouldn't react like that to a review.
But thank you for staying straight (and everyone else who did or does).
Hey @Rubben , do you feel left out?
Yes.
@jsg when you're reviewing providers, as a standard practice do you have a metric to go beyond one month's worth of benchmarking?
I'd like to, very much, but usually I benchmark only 24 to 72 hours. If it's my server I repeat it from time to time though.
Btw, running my benchmark for a full month or even longer would go beyond the traffic volume of many VPSs.
Instead of running a lot of benchmarks over 1 to 3 days, couldn't you run your benchmarks once every 6 or 8 hours for a couple of weeks? That would paint a more accurate picture. But to be fair, at least we know how the VPS performs when the node is loaded.
I could kind of. But running a benchmark requires supervision. Also my aim is to collect data over a full day (or two or three) that is, at day and night time.
But often I repeat my benchmark after a while and/or every couple of months. That however is only possible with servers paid for a longer time period (I tend to buy for a full year though). But I publish those additional results only if I see a significant change.
Since the title promises both — worth keeping in mind that a benchmark and a review are different things.
A benchmark is a point-in-time performance snapshot. A review requires time under real conditions — a quarter (subjective) of actual use, tracking stability, consistency, provider behavior. One without the other is half the picture.
IMHO, the review part is still TBD...
Friendly intentions? You used italics excessively. You do know you when reading italics you read it differently so you come across as a stupid teen girl with an attitude problem, right?
Your italics use is not normal.
Oof, you failed hard there.
I don't know why you thought he bought it two weeks ago. He only did a day of testing.
There's no point telling jsg this, he's been told since before he was "Resident Benchmarker".