Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!


How people create VPS with only 256mb ram?
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.

How people create VPS with only 256mb ram?

I ask my self: How people create VPS with only 256mb ram?
Lets say they get dedi with 64 GB ram and 24 core, buy /24 IPv4 subnet - 256 IP's, split the dedi for 256 VPS? its will give us only 24/256 core= less than 0.1 core for each :/
How its works?

Comments

  • why not 32MB?

  • you count on not everyone using all of their resources at once. its simple over selling. Airlines do it too. They normally over sell the seats and hope a few people won't show up. and if that doesn't happen they offer a "free" flight another time to people in exchange for being "bumped" to the next flight. but with a VPS its just hope you don't get abusers and if you do you ban the abusers

    Thanked by 1orel
  • @century1stop said:
    why not 32MB?

    i cant undersand what you are want from me, i ask how they do it, a lot of people here offer 256mb VPS...


    What is the minimum core that need to be for 256mb VPS (linux)?

    Thanked by 1century1stop
  • @orel said:
    What is the minimum core that need to be for 256mb VPS (linux)?

    1 shared core/thread

  • orelorel Member
    edited February 2016

    @masterqqq said:
    1 shared core/thread

    "shared core", between how much VPS? lets say i have 64GB ram & 24 core dedi, how much 256mb VPS can i create with that?

  • @AnthonySmith is probably the expert on this :)

  • VirMachVirMach Member, Patron Provider

    CPU always gets "oversold" unless you get some sort of "virtual dedicated" plan that guarantees you full usage. Let's go ahead and use your example real quick and explain it a little further. So we have a dedicated server with (for simplicity) 24 logical cores, 256 IP addresses - and we'll ignore other details.

    • There will not be only 256MB RAM plans on there: Some people may want more RAM, some people may need more IP addresses, so first we need to assume that there won't ever be 256 plans sharing CPU
    • There will be fair usage policy: Almost any provider will have some sort of acceptable usage level. If they give everyone 4 cores, they expect you to be able to burst to that amount at times but not always use the CPU
    • So here's what the real-world math looks like: Maybe something like 50 to 100 people on that node. If everyone gets a core, they'll be expected to use something like 25% average CPU.

    And this works out. It's a virtual environment. There's never going to realistically be a situation where there will (1) be 256 people on a node with all the same small budget plan AND (2) they all try to use all of their CPU at the same time AND (3) they're all permitted to use all of the CPU 24/7.


    Here's a quick real-world example on one of our new servers that has 24 logical cores.

    54 Virtual Servers

    59 IPv4 Addresses Used

    0.95 Load Average

    2.98% CPU Used

    6.85 GB RAM Used

    These are 128MB to 2GB plans, with 0.5 logical core to 1.5 logical core, or about 54 logical cores average. We only have 24 logical cores, but instead of (54/24) 225% being used, only about 3% of the CPU is being used.

    Thanked by 2orel mtsandeep
  • AnthonySmithAnthonySmith Member, Patron Provider

    @orel this is a how long is a piece of string question, there is no 1 correct answer because it is impossible to predict what customers demands will be.

    In some scenarios I could have 500 x 128mb containers running on a node with 8GB ram, single disk and an old quad core cpu with no problems at all, at other times 50 customers could render it unusable.

    because you cannot predict the use the only thing that will answer this is experience with a twist of 'take it as it comes'

    Regarding the CPU specific question, there are a number of methods, you can create cpu core pools e.g. if you have 8 logical cores you can split that in to 4 pools, each pool contains 2 cores, then you pin servers to a pool.

    You also add weighting to cpu time per vps so if you have 2 guests that need cpu time at the same time and they both have equal priority one will wait for the other to finish, iirc this is based on a per I/O request so that does not mean if 1 is using 100% the other will get 0%, cpu cycles are fast, this allows guest servers to share a cpu core without hurting each other to much, think of it like another level of hyper threading if it makes more sense to you that way.

    On container based systems it is a bit different, 1 server can screw everyone else so you need to define your limits carefully to ensure your host server has a greater share than any single container and no container can consume all resources.

    On avg though at any given time 90% of servers are idle in terms of CPU time.

    Thanked by 2orel miTgiB
  • KuJoeKuJoe Member, Host Rep

    Shared CPU + CPU cycle priority = VPS

    Thanked by 1AnthonySmith
  • AnthonySmith said: this is a how long is a piece of string question

    What a great phrase, will have to add it to my vocabulary

  • Let's rewrite the phrase so it's legible:

    This is a "how long is a piece of string" question.

    Thanked by 1ucxo
  • globalRegisters said: Let's rewrite the phrase so it's legible

    Don't be a hater ;)

  • +1 on the fair use policy.

    You won't have anybody using a full core unless they're doing something CPU intensive.

  • Common CPU spikes can happen on SSH scans/bruteforcing

  • AnthonySmithAnthonySmith Member, Patron Provider

    globalRegisters said: "how long is a piece of string"

    It is twice as long as it is from the middle to one end.

    Thanked by 2ucxo FalconsCloud
  • @miTgiB said:
    Don't be a hater ;)

    I wasn't trying to be negative. I just couldn't resist.

Sign In or Register to comment.