Score:-1

Server Response Time Fluctuation on VPS for Static Files

us flag

I monitored some of my VPS (shared CPU from major suppliers), and response time changes over time from 50ms to 2000ms. There is no traffic on the server and there is no database. So is this because of a shared CPU, or a shared network? Maybe the provider says they give us a 1Gbit connection but they can't give it on peak hours for all servers? How can I figure out whether it is a shared CPU or a shared network?

Score:1
cn flag

And the question is? Oh, you got a VIRTUAL server running on a hardware you do NOT control and there is NO way for you to measure whether or not the hardware has any load on it, or load spikes.

Here is what you do not do: GO to an internet forum with no hardware level information.

here is what you do: contact your "major supplier" support - they are there for that, you pay them for that and - they have access to, you know, the hardware.

Maybe the provider says they give us a 1Gbit connection but they can't give it on peak hours for all servers?

MAYBE you read what you signed up for. MAYBE you get some qualification before going to a place for professionals.

  • The 1GB is port speed, not internet speed.
  • No provider in the world will not oversell. You want 1GB to the internet, you pay for it - explicitly. Which will be a little more expensive (like hundreds of times) than your VPS.

So no, they do not "maybe" - they deliver exactly what you paid for: 1gigabit per sever to their backbone. Note the per server - that is very unlikely for your VPS (because only so many physical ports per server and cables cost money) but per physical server.

But you are also more likely that the storage subsystem just has IO spikes. We will not be able to answer because you provide no and have no access to any metrics from the hardware layer.

Off to support. They know.

doraemon avatar
us flag
So shortly, you are saying there is no way to detect the root cause on a VPS and you are suggesting a dedicated network with a dedicated server for consistent results? Or is this the same for VMWare-based VPS services too where they give dedicated CPU and network connection?
cn flag
"there is no way to detect the root cause on a VPS" Oh, there is - I do that all the time. I log in on the hardware console. There is no way to do it FROM INSIDE THE VPS. You are a passenger; you do not see the maintenance log of the taxi driver. "Or is this the same for VMWare-based VPS services too where they give dedicated CPU and network connection?" - oh, another one. CPU and network, but you STILL share memory bus and storage IOPS budgets, you know. Someone else running an AI LLM on the CPU and it eats ups as much CPU bandwidth as it can get.
I sit in a Tesla and translated this thread with Ai:

mangohost

Post an answer

Most people don’t grasp that asking a lot of questions unlocks learning and improves interpersonal bonding. In Alison’s studies, for example, though people could accurately recall how many questions had been asked in their conversations, they didn’t intuit the link between questions and liking. Across four studies, in which participants were engaged in conversations themselves or read transcripts of others’ conversations, people tended not to realize that question asking would influence—or had influenced—the level of amity between the conversationalists.