How exactly did you get 10,000 accounts (offline mode?), or a method to place them all onto the server?
And that's a horrible test anyways, you could fit hundreds of thousands of people when all they're doing is standing still in spawn.
They were all randomly walking all around the world. I did not use 10,000 actual accounts but rather modified the server to create players from the server (so no client emulation was used)
I used to have thousands of account credentials before migration was introduced :-P
And no, I did not obtain them myself... they were leaked on a huge forum database that required people to use their Minecraft username/password to synchronize with their forum ;-)
There's no way you had 10,000 people randomly walking around on that server. The CPU would not be able to handle the load.
Rollback Post to RevisionRollBack
https://allgamer.net/ - Minecraft, KVM VPS, and Dedicated Server Hosting
Email/MSN: clinton -at- allgamer -dot- net
AIM: Clinton4664 | Skype: clinton_collins I am a representative of AllGamer
There's no way you had 10,000 people randomly walking around on that server. The CPU would not be able to handle the load.
My two CPUs were able to handle 10,000 players just fine. I think your outlook is too narrow in the subjective matter and therefore can only provide so much argumental evidence where such is relative to the general.
The real answer is 2480 slots
This is the maximum Minecraft will hold.
However, most computers on the market could never ever hold more than 1000 slots
I've see a 700-slot server.
But still, these computers are tens of thousands of dollars, so the maximum you will usually see is 350 slots
These are very common.
This isn't true in the current version. The client will reject any slot value higher than 2480.
I cannot possibly see the maximum slot value being 2480. Perhaps you meant to say 2048? In that case it would be up to slot 2047, 2048 slots in total. Anyways regardless of this I did not connect with a client to the server but instead traced the time it took to complete a cycle.
To resolve this issue you can simply not update players who you cannot see (and maybe even players that aren't in your render distance) and say even set a max to that, being maybe 256?
Computers that can handle 1000 slots aren't that expensive. I would use a machine with similar hardware to my own to host an 1000 slot server. Note that my machine only costed $1800 used.
I cannot possibly see the maximum slot value being 2480. Perhaps you meant to say 2048? In that case it would be up to slot 2047, 2048 slots in total. Anyways regardless of this I did not connect with a client to the server but instead traced the time it took to complete a cycle.
To resolve this issue you can simply not update players who you cannot see (and maybe even players that aren't in your render distance) and say even set a max to that, being maybe 256?
Computers that can handle 1000 slots aren't that expensive. I would use a machine with similar hardware to my own to host an 1000 slot server. Note that my machine only costed $1800 used.
Its about the maximum the server software may handle, regardless of hardware. You need to re-run your expirement with 10k real clients for accurate results.
My two CPUs were able to handle 10,000 players just fine. I think your outlook is too narrow in the subjective matter and therefore can only provide so much argumental evidence where such is relative to the general.
10,000 "players".
Either try testing it with 10,000 real clients, or stop pretending you can hold 10,000 players.
Computers that can handle 1000 slots aren't that expensive. I would use a machine with similar hardware to my own to host an 1000 slot server. Note that my machine only costed $1800 used.
"only $1800"
Last time I checked, $1800 isn't cheap. And good luck running 1000 slots on a single server.
“Two things are infinite: the universe and human stupidity; and I'm not sure about the universe.” — Albert Einstein
"Never try to teach a pig to sing; it wastes your time and it annoys the pig." — Robert Heinlein
Either try testing it with 10,000 real clients, or stop pretending you can hold 10,000 players.
All behavior that is serious to cycle time is ran regardless of using a networked connection by 10,000 users or by instantiated players on the server so long as the instantiated players perform expected behaviors by real clients.
Computers that can handle 1000 slots aren't that expensive. I would use a machine with similar hardware to my own to host an 1000 slot server. Note that my machine only costed $1800 used.
That, considering my subjective reference being calibrated from souljabri557, is not relative to the general but directly relative to the reference.
All behavior that is serious to cycle time is ran regardless of using a networked connection by 10,000 users or by instantiated players on the server so long as the instantiated players perform expected behaviors by real clients.
No, that's not true, not unless each "player" is a completely separate standalone Minecraft process that performs random actions in the fashion a normal player would. Your "emulation" isn't a good benchmark.
“Two things are infinite: the universe and human stupidity; and I'm not sure about the universe.” — Albert Einstein
"Never try to teach a pig to sing; it wastes your time and it annoys the pig." — Robert Heinlein
The highest I've seen was MineZ right after Beta when their servers had a 165-person cap and were almost always full. Since Minecraft server is single-core, there is a definite limit. Connection speed, processor clock speed (per core, since it is single-core only), allocated RAM, and hard drive speed are the biggest factors.
No, that's not true, not unless each "player" is a completely separate standalone Minecraft process that performs random actions in the fashion a normal player would. Your "emulation" isn't a good benchmark.
$1800 isn't cheap, nor will it buy you 10,000 slots.
All behavior that is serious to cycle time is ran regardless of using a networked connection by 10,000 users or by instantiated players on the server so long as the instantiated players perform expected behaviors by real clients.
Do you dap read my text? In all due respect, this is the third time I have seen the dismissal of my post direction.
Is it that you not see the directional premise of my posts?
[/background][/size][/font][/color] That, considering my subjective reference being calibrated from souljabri557, is not relative to the general but directly relative to the reference.[/background][/size][/font][/color]
Your stand along the reputable parchment by which you do present here, is also such at a gate of decline could be mentioned to such evidently profound dogmatism. For one to also be a conclude of such state, who holds and writes against his own empirical approach to my latter premise, also does conclude to himself, that he by himself is alone in argument, for the argument of such at no means of understanding could never be concluded sound.
Your stand along the reputable parchment by which you do present here, is also such at a gate of decline could be mentioned to such evidently profound dogmatism. For one to also be a conclude of such state, who holds and writes against his own empirical approach to my latter premise, also does conclude to himself, that he by himself is alone in argument, for the argument of such at no means of understanding could never be concluded sound.
This is highly exaggerated and somewhat sarcastic. Please retype your statement or edit your post to something that is worth replying to.
As three of us have pointed out to you, your test environment is not an accurate simulation of the loads 10,000 players would place upon a server.
Consider this: At an absolute minimum ram footprint of 20MB per player, and leaving nothing for everything else, that's 195GB of RAM being used for these 10k clients. Going off of a baseline network usage of 100Kbps per player, that's a full gigabit of network usage.
We've already established that the network requirements are at the extreme upper edge of what a single gigabit interface could hold, and the RAM requirements are at a level that requires extremely expensive and high-end Xeon servers to supply. This is completely ignoring the CPU requirements for 10,000 users, not to mention the extreme disk IO you'd be seeing from this test.
“Two things are infinite: the universe and human stupidity; and I'm not sure about the universe.” — Albert Einstein
"Never try to teach a pig to sing; it wastes your time and it annoys the pig." — Robert Heinlein
As three of us have pointed out to you, your test environment is not an accurate simulation of the loads 10,000 players would place upon a server.
Consider this: At an absolute minimum ram footprint of 20MB per player, and leaving nothing for everything else, that's 195GB of RAM being used for these 10k clients. Going off of a baseline network usage of 100Kbps per player, that's a full gigabit of network usage.
We've already established that the network requirements are at the extreme upper edge of what a single gigabit interface could hold, and the RAM requirements are at a level that requires extremely expensive and high-end Xeon servers to supply. This is completely ignoring the CPU requirements for 10,000 users, not to mention the extreme disk IO you'd be seeing from this test.
Your "results" are invalid.
A Minecraft player instance isn't anywhere near 20MB. Perhaps 20KB... as a maxima.
Also, I find it very frustrating that you do not take my posts into consideration. All behavior that is serious to cycle time is ran regardless of using a networked connection by 10,000 users or by instantiated players on the server so long as the instantiated players perform expected behaviors by real clients.
As three of us have pointed out to you, your test environment is not an accurate simulation of the loads 10,000 players would place upon a server.
Consider this: At an absolute minimum ram footprint of 20MB per player, and leaving nothing for everything else, that's 195GB of RAM being used for these 10k clients. Going off of a baseline network usage of 100Kbps per player, that's a full gigabit of network usage.
We've already established that the network requirements are at the extreme upper edge of what a single gigabit interface could hold, and the RAM requirements are at a level that requires extremely expensive and high-end Xeon servers to supply. This is completely ignoring the CPU requirements for 10,000 users, not to mention the extreme disk IO you'd be seeing from this test.
Your "results" are invalid.
It's not worth arguing. In the end there are some of us that know the limits of minecraft servers.
Rollback Post to RevisionRollBack
https://allgamer.net/ - Minecraft, KVM VPS, and Dedicated Server Hosting
Email/MSN: clinton -at- allgamer -dot- net
AIM: Clinton4664 | Skype: clinton_collins I am a representative of AllGamer
A Minecraft player instance isn't anywhere near 20MB. Perhaps 20KB... as a maxima.
Also, I find it very frustrating that you do not take my posts into consideration. All behavior that is serious to cycle time is ran regardless of using a networked connection by 10,000 users or by instantiated players on the server so long as the instantiated players perform expected behaviors by real clients.
Did.. Did you just say kilobytes? Twenty.. Twenty KILOBYTES?
Okay, that's it, I'm outta here folks. I'll go join Clinton in greener pastures.
“Two things are infinite: the universe and human stupidity; and I'm not sure about the universe.” — Albert Einstein
"Never try to teach a pig to sing; it wastes your time and it annoys the pig." — Robert Heinlein
I used to have thousands of account credentials before migration was introduced :-P
And no, I did not obtain them myself... they were leaked on a huge forum database that required people to use their Minecraft username/password to synchronize with their forum ;-)
Just a developer :-)
https://allgamer.net/ - Minecraft, KVM VPS, and Dedicated Server Hosting
Email/MSN: clinton -at- allgamer -dot- net
AIM: Clinton4664 | Skype: clinton_collins
I am a representative of AllGamer
Just a developer :-)
Exactly. It is the software as the ultimate limitation, not the hardware.
To resolve this issue you can simply not update players who you cannot see (and maybe even players that aren't in your render distance) and say even set a max to that, being maybe 256?
Computers that can handle 1000 slots aren't that expensive. I would use a machine with similar hardware to my own to host an 1000 slot server. Note that my machine only costed $1800 used.
Just a developer :-)
Its about the maximum the server software may handle, regardless of hardware. You need to re-run your expirement with 10k real clients for accurate results.
10,000 "players".
Either try testing it with 10,000 real clients, or stop pretending you can hold 10,000 players.
I set my slots to 2481 and connected, lo and behold, it worked just fine!
"only $1800"
Last time I checked, $1800 isn't cheap. And good luck running 1000 slots on a single server.
"Never try to teach a pig to sing; it wastes your time and it annoys the pig." — Robert Heinlein
Do you dap read my text? In all due respect, this is the third time I have seen the dismissal of my post direction.
Is it that you not see the directional premise of my posts?
That, considering my subjective reference being calibrated from souljabri557, is not relative to the general but directly relative to the reference.
Just a developer :-)
No, that's not true, not unless each "player" is a completely separate standalone Minecraft process that performs random actions in the fashion a normal player would. Your "emulation" isn't a good benchmark.
$1800 isn't cheap, nor will it buy you 10,000 slots.
"Never try to teach a pig to sing; it wastes your time and it annoys the pig." — Robert Heinlein
Just a developer :-)
He is reading fine, as am I. You are not properly emulating the environment of 10,000 clients.
Your stand along the reputable parchment by which you do present here, is also such at a gate of decline could be mentioned to such evidently profound dogmatism. For one to also be a conclude of such state, who holds and writes against his own empirical approach to my latter premise, also does conclude to himself, that he by himself is alone in argument, for the argument of such at no means of understanding could never be concluded sound.
Just a developer :-)
This is highly exaggerated and somewhat sarcastic. Please retype your statement or edit your post to something that is worth replying to.
Just a developer :-)
As three of us have pointed out to you, your test environment is not an accurate simulation of the loads 10,000 players would place upon a server.
Consider this: At an absolute minimum ram footprint of 20MB per player, and leaving nothing for everything else, that's 195GB of RAM being used for these 10k clients. Going off of a baseline network usage of 100Kbps per player, that's a full gigabit of network usage.
We've already established that the network requirements are at the extreme upper edge of what a single gigabit interface could hold, and the RAM requirements are at a level that requires extremely expensive and high-end Xeon servers to supply. This is completely ignoring the CPU requirements for 10,000 users, not to mention the extreme disk IO you'd be seeing from this test.
Your "results" are invalid.
"Never try to teach a pig to sing; it wastes your time and it annoys the pig." — Robert Heinlein
Also, I find it very frustrating that you do not take my posts into consideration. All behavior that is serious to cycle time is ran regardless of using a networked connection by 10,000 users or by instantiated players on the server so long as the instantiated players perform expected behaviors by real clients.
Just a developer :-)
It's not worth arguing. In the end there are some of us that know the limits of minecraft servers.
https://allgamer.net/ - Minecraft, KVM VPS, and Dedicated Server Hosting
Email/MSN: clinton -at- allgamer -dot- net
AIM: Clinton4664 | Skype: clinton_collins
I am a representative of AllGamer
Did.. Did you just say kilobytes? Twenty.. Twenty KILOBYTES?
Okay, that's it, I'm outta here folks. I'll go join Clinton in greener pastures.
"Never try to teach a pig to sing; it wastes your time and it annoys the pig." — Robert Heinlein