New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
I get your point but: different story. My benchmarks aim to provide a (fast based) impression of what potential customers can expect
I find your idea interesting, but measure against what? Unless providers gave concrete info like "vCore(s) dedicated x%, disk dedicated y%, network dedicated z%" measurements following your idea would have very little meaning.
Plus there is the other, and potentially very ugly, factor: if one user pushes all resources to the limits, then (a) other users will suffer, and (b) the "benchmarker" will be thrown out except with some very pricey products (basically full VDS) or some utterly careless providers tolerating what basically is abuse.
FWIW, my benchmarks do push the VPS, albeit in a civilized manner. Explanation: all the tests run in "slices", so e.g. the processor and memory tests seem to be one test but actually are thousands off "slices" (like e.g. 2 GB test actually is x thousand "sub tests" of say 100 KB) with small pauses (can be configured, default is 5 ms) in between to allow the neighbour VPSs on the node to run too.
Now, 5 ms may sound like a lot but actually all in all my benchmark pushes the test VPS harder than 95% of all real usage (by e.g. DB + web server).
IMO the real solution for the problem you're addressing would require providers to be more open and honest ... which is why it's very unlikely. You see, when a provider puts 2 or 3 customers on a hardware thread they could say "50% (or 33%) dedicated" - but if it's 5 or even more customers that is, if a vCore actually is 20% or even less of a hardware thread those providers are very unlikely to play with open cards. And I guess that's particularly relevant here at the low end of the market.
That said, one must also see that those providers are not totally off when considering that the vast majority of VPSs actually are way beyond what is actually needed or, in other words, the vast majority of us buy way more than what we actually need.
Practical honest example: myself and my name servers. My master runs on a multi GB multi-core higher end VPS from an excellent provider - which honestly is ridiculous. And so is my half dozen secondaries running on diverse mini VPS (typ. 1 vCore, 256 MB or 512 MB, 7+ GB disk, usually SSD). That is utterly ridiculous too because even at peak times those (tiny, mind you) VPSs hardly ever reach a load of even just 0.1 (allowing a cheap provider cramming plenty of those tightly onto a single HWT/vCore).
That said, my benchmark software does allow for "go all in without mercy" benchmarking and I occasionally use that mode when benchmarking (e.g. the Contabo) VDSs. But not when testing VPSs; it simply doesn't make sense IMO, it's more meaningful to provide numbers along the line of "pushing the VPS relatively hard but not beyond the load of 95+% of potential customers.
What kinds of tests do you run for the CPU benchmark? Real world algorithms similar to Geekbench?
Wow.. that is one hell of text wall..
What is "real-world"? OpenSSL?
I do run similar tests, yes, but way more optimized for benchmarking a system than just some calls into third party libraries.
My benchmark is, to the best of my knowledge, also the only benchmark with true fully random read and write (disk) testing that is, both the data and the sector are fully and true (high quality practrand tested) random.
Maybe you'll find some benchmark/review on twitter (or are even a couple of hundred letters too much for you to read? If so, sorry, I don't know any benchmark comics).
Geekbench has documentation which talks about the algorithms they use. But in general they run HTML parsing and DOM traversal, Dijkstra, AES encryption/decryption, and some image processing algorithms.
gcore labs recently added Romania, pretty cheap too.
I'm confused because @Calin isn't a provider. Or did he lose his tag?
(misunderstanding/post I responded to has changed)
Sorry, I edited my post, I meant Calin, not you. I had too many beers and mixed the threads.
@Arkas I temporarily lost the tag, I will fix all the problems at the site that were brought to me when the tag was revoked
I think the Hosthatch performance was surprising, for the price. As a fellow hostsolutions refugee, I ended up at BuyVM in Luxembourg. I noticed all your servers don't have much HD space, but I'm assuming you're looking for speed and not space. Thank you for the benchmarks.
I like the concept and your reviews look thorough, but HostSolutions' fall was due to storage units, and as I have seen, none of these are high storage deals. For me, the most important metric was price per terabyte where he smashed every other provider (granted, at what price?).
The topic literally made me think there was some kind of terrorist attack.
Not really, I myself also have a couple of systems with larger storage cap. It just so happens that cheap or very cheap VPSs tend to come with rather small disk cap.
You're welcome. It makes me happy to see that my work is useful for some others.
I get your point but IMO the storage systems failure was the trigger but not the reason for HS' fall down. That was the owners utter ignorance, almost total lack of communication (plus what he said being pretty much worthless) and, so I guess, his psycho derangement.
As for the price of his storage boxes, sorry, but that must be calculated not only in $$ paid but also in data lost, time lost, efforts wasted, etc. All in all his storage stuff turned out to be ridiculously expensive.
Everyone to his liking but AFAIC I've learned the lesson and prefer to pay some more $$ with a reliable, good quality, and trustworthy provider like NexusBytes / @seriesn ... and to sail better and cheaper at the end, when all factors are added up.
Side note: I wonder whether (and if not then why) @dedicatserver_ro offers large storage VPS, say 0.5 to 10 TB, because while I don't like his (overly) complex (kind of) NAS for normal VPS I think it might serve really well for storage VPSs.
No one from LET provider uses NVMe-oF over RDMA - RoCE v2 or BeeGFS over RDMA - RoCE v2 , because of knowledge and costs.( RDMA - RoCE v2 is used in Supercomputers for example Oak Summit - Rank 2 in the world , use BeeGFS ), nobody cares
if you haven't noticed on the site you can add on each VPS or dedicated server: optional up to 300TB Bloc Storage optional 0,01 € per GB/month, minim 500GB ( need to pay for )
In my opinion :
PS. more than half of the new customers in the last 3 month we closed for non-compliance with TOS!!
MivoCloud are Moldavian, isn't? Nation are very closest but not exactly.
Only difference.. they join Soviet Union, you join Hitler's Germany 80 years ago.
Me not care whether they are romanian or moldavian. If they want me to review them, I'll do it.
2 equally crappy outcomes...
America and Britain join Soviet Union 80 years ago.. 2 crappy outcomes?
No, they didn't join the Soviets, they were allied with them for the duration of the war.
The second the war was over, they pointed their guns at each other.
I don't know why people don't just time their install scripts for their particular use cases to know which server works best for them.
I mean join forces, not a being part of country of course. Romania also join Hitler's forces but fail quick. But Moldova was being a part of Soviet Union.
Thanks to the victory of the united parts, you now exist.
I'm German.
@jenkki
@Arkas
https://en.wikipedia.org/wiki/History_of_Moldova
the whole of Moldova, half in Romania.
Your uncle Fuhler fail at his time. With other satellites such a Romania. That we are talking about. Moldavia a part of winners. Like a other Soviet Union nations.