What is your comment about this benchmark?

ironhide

http://serverbear.com/benchmark/2014/02/05/iDHwWhJUgNDxQXfK

please do share what do you think

myhken

A OK UnixBench result, but not so good IO speeds (DD around 941 kB/s, thats really low), and the networks speeds is not very good at all for all locations.

If it was any of my servers, I have asked for a refund, or just stopped using them.

kyaky

looks like it's an openvz with 8 cores from the data. 3064 bench is not that bad. but I've seen weloveservers' 4 cores do 3200+ and their 8 cores do 5600+ bench.

I/O could be better.

It's quite reasonable for the spec if it's cheap enough.

MikeIn

nothing very abnormal but you have access to all 8 cores, nice

Bandwidth is not very nice.

SpeedBus

Hey,

Just adding in, I/O will improve once we move client's over to a newer node with a better RAID card this week/early next week. Emails regarding this will be sent out in a day or 2 once we have a confirmed date/exact date at hand. Though what surprises me is the network speeds reported there, I ran a few tests off https://www.linode.com/speedtest/ and I'm easily able to go upto at LEAST 4 MB/s on all the locations (least being Japan followed by London and highest being Fremont)

EDIT: basically, give it a second go once we have moved you over to the new server, should be far better then.

SpeedBus

myhken said: (DD around 941 kB/s, thats really low)

That's the dsync command which is showing you the low speed, ie,

dd if=/dev/zero of=sb-io-test bs=64k count=16k oflag=dsync

I don't think that's the "standard" command for DD tests, from what I've seen most scripts rely on the following command,

dd if=/dev/zero of=sb-io-test bs=64k count=16k conv=fdatasync

myhken

SpeedBus said: SpeedBus

Still, I get this result on my Ramnode SSD-cached node:
[root@xxxxx ~]# dd if=/dev/zero of=sb-io-test bs=64k count=16k oflag=dsync 16384+0 records in 16384+0 records out 1073741824 bytes (1.1 GB) copied, 10.8514 s, 98.9 MB/s

Thats around 100 times better then your server.

SpeedBus

myhken said: SSD-cached node

That's why it's 100 times better We don't use any SSD cache on our nodes at the moment

EDIT: let me explain a bit more,
As you see on, https://romanrm.net/dd-benchmark

dd bs=1M count=256 if=/dev/zero of=test oflag=dsync 
Here dd will ask for completely synchronous output to disk, i.e. ensure that its write requests don’t even return until the submitted data is on disk. In the above example, this will mean sync'ing once per megabyte, or 128 times in total. It would probably be the slowest mode, as the write cache is basically unused at all in this case.

It would probably be the slowest mode, as the write cache is basically unused at all in this case.

From what I understand,

Although the test ignores the write cache on the RAID card, but since the VPS you tested on has a SSD cache, it will surely get you good results since the SSD acts in as a cache making up for the onboard RAID card cache which is ignored by this test We are looking into SSD caching for our future machines since it has many advantages as I showed above.

Again, I maybe wrong, but anyways I'm pretty sure the new server we're getting in will be far better than the current server

myhken

@SpeedBus said:

My iwStack.com node is 34 times faster then your server, thats not SSD cached or SSD.

[root@xxxxx ~]# dd if=/dev/zero of=sb-io-test bs=64k count=16k oflag=dsync 16384+0 records in 16384+0 records out 1073741824 bytes (1.1 GB) copied, 31.3175 s, 34.3 MB/s

My Backupsy.com node is 55 times faster then your server, thats not SSD cached or SSD

[root@xxxxx ~]# dd if=/dev/zero of=sb-io-test bs=64k count=16k oflag=dsync 16384+0 records in 16384+0 records out 1073741824 bytes (1.1 GB) copied, 19.3338 s, 55.5 MB/s

My INIZ node is 174 times faster then your server, thats not SSD cached or SSD
[root@xxxxx ~]# dd if=/dev/zero of=sb-io-test bs=64k count=16k oflag=dsync 16384+0 records in 16384+0 records out 1073741824 bytes (1.1 GB) copied, 6.16311 s, 174 MB/s

And the list continues....just making a point.

SpeedBus

Yep, I get the point, hence the reason we've ordered a new server with a better RAID card already to take care of this issue

EDIT: Sorry if I sounded rude, been a long weekend+Monday+Tuesday+Today aand lack of sleep whilst getting rid off the abusers we got over the weekend on that server. I've got some dates in for the new server coming in, and am drafting the email with more info which will be sent to clients once I have the exact server delivery date.

EDIT: Email sent to all clients regarding new server and time of migration etc.