Howdy, Stranger!

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


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.

Black Friday strategy: one bigger VPS, or a few small ones?

2»

Comments

  • I used to have lots of small VPSes* but I've been steadily sticking everything onto two VDSes instead. I've come to prefer consistently reliable servers.

    * I still do, they're just idlers now.

  • This is great but please please please for the love of god and all the saints - make backups! Just don't ask any questions, make backups. And keep them off that server, which you are backing up.

  • Bots will win, I don't expect too much

  • @JohnFilch123 said:
    This is great but please please please for the love of god and all the saints - make backups! Just don't ask any questions, make backups. And keep them off that server, which you are backing up.

    "I back up to the cloud regularly, but for small tests it's fine; once it's frequent, they will suspend you."

  • @nikio said:
    @ OP - your post reads like Grok generated it. If that is the case please just ask Grok for his opinion since he already gave you two options.

    In any event keep in mind that not all VMs are created equal. A single core on a 9950x may be better than four cores on a Xeon v1 shitbox; at the same time that core might be so heavily throttled that an Atom C2750 actually performs better.

    Lastly you're asking people on a forum who have dashboard-measuring contests for idlers.

    "Brother, I finally found it. It is not easy at all to find someone of the same kind." >:) >:)

  • yoursunnyyoursunny Member, IPv6 Advocate

    We buy many small VPS across many providers, to achieve geo-diversity and worldwide coverage.
    Then we buy a handful of large VPS across several reputable providers, for more demanding services like llama.cpp.

    Whether a service is “production” or “experiment” doesn’t affect whether it’s going to a large machine or a small machine.
    Placement is determined by the resources required to deploy this service.
    We obviously consider the yoursunny.com website as “production”, yet it goes into a 1C4G20G instead of a 2C8G100G VPS.

    We reduce the blast radius of a “learning mistake” by:

    • Keep detailed records on the installation procedure, so that it can be repeated. The procedure is then continuously refined when the repeat installation is carried out.
    • Manually install each KVM host from ISO, so that every installation starts with a known baseline.
    • Conduct most experiments inside LXC containers rather than on the KVM host, so that the container can be deleted and re-created.
    • If a procedure requires editing the KVM host directly, conduct the experiment on a spare machine first.

      • If this involves an upgrade of existing machine, first install that spare machine with the old version following the detailed notes, then try the upgrade.
      • Keep detailed notes on the procedure so it can be applied to more important machines.
    • Set aside some spare time for disaster recovery before committing to a risky procedure on an important machine.

    Finally, keep backups of the detailed notes, and backups of any service state needed for restoration.
    Any VPS can be wiped by mistake; any provider can deadpool; you can be in the ICU or the county jail with the hosting invoice is due.
    Have a plan on what to do with each deployed service when that happens.
    See also: How Many Servers Do You Need for a Crucial Website?

    Thanked by 2pinwei 0xC7
  • Wow, your AI-auto-bots sure can talk, the longest message in this thread. Is it running on Deep-Reek?

  • @nikio said:

    @cold said:
    no VPS, new sneakers and sex toys, they bring more joy and won't renew the prices to a higher one next year

    Unless the sex toys are made by LG, internet-connected, spy on you and require a subscription to use. And right in the middle of the action Sam Altman's voice comes on and says 'Hey I needed that RAM"

    Why LG not Samsung

  • coldcold Member

    @nikio said:

    @cold said:
    no VPS, new sneakers and sex toys, they bring more joy and won't renew the prices to a higher one next year

    Unless the sex toys are made by LG, internet-connected, spy on you and require a subscription to use. And right in the middle of the action Sam Altman's voice comes on and says 'Hey I needed that RAM"

    all are made out of BIO wood ! im safe bro

  • I'd lean toward a few small ones for your use case, mainly because it keeps the learning box separate from the personal site. If I break something while experimenting, I don't want it taking down the site too.

    I'd still aim for at least 2GB on the main one — 1GB can get tight once you add a database and a couple of services. Docker on the bigger box makes sense for keeping projects separated without managing multiple servers, but I'd keep the learning box as a plain separate VPS so I can rebuild it freely.

    For timing, I'd watch for a good deal on the main box first and only add the learning box when there's a cheap offer, rather than buying both at once.

  • YachiyoYachiyo Member
    edited 1:06AM

    something like 3–4 cores / 6GB RAM

    Anything with fewer than 8 vCores or less than 16GB RAM doesn't qualify as "big."

    ...Unless it is a dedicated server

  • truemagictruemagic Member

    I've five active and dozens of smaller idling VPSes. I've two "big" ones and three small ones (costing $7/y).

    Before some providers increased their prices, I was going to give up on several $6/y VPSes, but now I'm having second thoughts about renewing them for just another year. Who knows, they might turn out to be worth keeping in the long run?

  • q2019q2019 Member
    edited 2:41AM

    I prefer bigger VPS.
    The more small one you have,the more time you need to maintain them,especially when you need to install security updates of vps(there are so many CVE this year due to the develop of AI,AI has helped discover a lot of vulnerabilities).
    And when a programe need burst resource, A bigger one may satisfy,but small one maybe OOM or CPU overload.

  • labraxlabrax Member

    not much I can expect from black friday for APAC location, everything under 24 usd a year is already good enough

Sign In or Register to comment.