1

VPS specs:

Linode 512 (more info)

Dedicated server specs:

Dual Atom (more info)

john -test on the Dedicated server

john -test on the Linode

jamm1ng
  • 47
  • 3

2 Answers2

6

Comparing Linode's Dual Quad-core Xeon CPUs vs your dual Atom D510 is like comparing a Lamborghini with a mobility scooter.

The Atoms have a tiny percentage of the processing power of modern CPUs. In fact, their performance is even poor compared to CPUs from 6 years ago. They are not built for grunt, but for small low-power devices such as "netbooks".

You will clearly get much, better performance from your VPS.

Now you may be thinking, you have to share the VPS CPU utilisation with dozens of other customers.

Well, even though you may be sharing your server with, say, 31 other customers, doesn't mean that you are getting 1/32 of the processor's cycles. That would only be true if all customers used 100% CPU the whole time, which isn't going to be the case. For most users, CPU usage will come in bursts, and most CPUs remain idle most of the time unless they are doing some sort of ongoing processing.

That said, I have to echo other people's comments that a benchmark like this isn't an indicator of real-world performance, especially because many tasks a server does are IO-bound rather than CPU-bound. So the speed of disk storage (IOPS) matters.

thomasrutter
  • 2,437
  • 1
  • 25
  • 34
2

It wouldn't be too surprising given that the CPUs (and therefore the surrounding hardware also) are nowhere near equal. The Atom CPU is tailored towards low utilization, low-power-consumption usage, and lacks a fair number of features in order to reduce its power draw.

Your specs show a number of factors straight away - The VPS has higher clock speed, larger memory bus, MUCH larger cache.

Chris Thorpe
  • 9,903
  • 22
  • 32
  • I'm a newbie, can you tell me if there is a HUGE difference in the results I got on the benchmark I ran? – jamm1ng May 03 '11 at 10:14
  • Well, quick calculations say the Linode is between 30 and 50% faster, depending on the test. It depends on your usage if this difference matters or is irrelevant. But what is your problem anyway? What do you want to do? Why are you using a password cracker as a benchmark tool? – Sven May 03 '11 at 10:39
  • I was reading some posts in a forum while searching 'ubuntu benchmark'. john was suggested there as a benchmark. The reason for the test are because I just bought this Dedicated server to replace one of my Linodes as it provides unmetered bandwidth. It seemed slower, so I wanted to compare it to the Linode. Although bandwidth was my main concern with the Linode, the CPU/Memory usage was reaching the limit of my Linode as well, so I'm not sure if I should move my site to the dedicated server or not considering these results. – jamm1ng May 03 '11 at 10:44
  • Well, if CPU was a concern on the Linode, your won't be happy with an Atom-based dedicated server. As I said, the Atom's results on a nearly exclusively CPU-bound test are between 30 and 50% of the Linode (I got that wrong earlier, the Linode is between 50% and 70% faster). Real-world differences might be quite different though, since few applications have the CPU as the bottleneck, memory and disk speed are often much more important. – Sven May 03 '11 at 10:51
  • @SvenW super disappointing. By the way the Linode in question was being tested with multiple services running on it (including the site in question) while the dedicated server was a fresh install with nothing on it. – jamm1ng May 03 '11 at 11:21
  • Most important thing for you to know is where your real bottlenecks are. I could imagine that you run out of memory relatively fast if you have only 512MB as in a Linode 512, so doubling the RAM might gain you much more than a tenfold increase in CPU speed in some cases. As I said, CPU is usually the least important factor compared to memory size and disk speed. – Sven May 03 '11 at 11:31
  • My worst bottleneck is bandwidth, I have an image sharing website (with more viewers than uploaders), so Bandwidth is number 1, after that, I/O and CPU. But from the tests, it looks like I have no point of moving the site as the CPU/IO will be significantly impacted on the Dedicated server. – jamm1ng May 03 '11 at 11:34
  • Are you currently maxing out your CPU and IO on the VPS? If you're only currently using a small portion of the CPu in the VPS, then your dedicated server may allow you to push more traffic up to the point where the CPU saturates. – Chris Thorpe May 03 '11 at 12:29
  • At peak hours I'm either maxing it out or very close to. How can I decide which is better? specifically for my needs that is. – jamm1ng May 03 '11 at 13:23
  • 1
    Load testing. That is, proper load testing with your actual workload loaded onto the dedicated server, then run stress tests to simulate your users. From that, you can decide which is best. – Chris Thorpe May 03 '11 at 13:40