M
M
Multigame2015-05-11 13:54:33
System administration
Multigame, 2015-05-11 13:54:33

What IO speed to expect from a dedicated server?

Good afternoon.
What speed to expect from a dedicated server?
I wanted to move from vds to a dedicated server, because of the low (as I thought 20mb / s low speed) speed of reading and writing from disk.
I asked for a dedicated server for testing (C2758, 8 GB, 2 * 1000 SATA (Raid 1), centos 6.5 32-64), but the speed of disk operations on it did not please (80 mb / s ext3, 130 mb / s ext4).
(tested like this: dd if=/dev/zero of=tfl count=512 bs=10240k conv=fdatasync)
UPD everything is the same, but without RAID - ext3 (104mb/s), ext4(140mb/s)
Question. Am I asking too much? (Expected 200 or more)
Or is it a server/setup issue? How much to expect from a dedicated server with two non-ssd and non-high-speed disks?

Answer the question

In order to leave comments, you need to log in

6 answer(s)
M
Maxim Grechushnikov, 2015-05-11
@maxyc_webber

hetzner. ubuntu 14.
dd if=/dev/zero of=tfl count=512 bs=10240k conv=fdatasync
512+0 records in
512+0 records out
5368709120 bytes (5.4 GB) copied, 14.5821 s, 368 MB/s

D
Denis Verbin, 2015-05-11
@rez0n

Most likely - you want too much.
PS. Have you looked at how many hours these disks have worked for you already?
P.S.S. Via smart (smartctl -d ata -a -i /dev/sda)

M
Melkij, 2015-05-11
@melkij

with two non-ssd and non-high-speed disks?

With one. RAID1 is a mirror. Write performance equals that of the slowest drive in the array.
I haven't seen tests of 7200rpm disks for a long time, yes, on such a synthetic load, but 200mb / s sequential write is quite a lot.

T
tzps, 2015-05-11
@tzps

In server business, megabytes / sec, etc. are not very indicative, Especially in sequential patterns.
Try measuring iops.
ps 140MB in sequential reading - not the ultimate dream, but not the bottom. Quite a normal figure in your configuration. But still, measure the IOPS if you plan to use the server multi-threaded.

S
Sergey Fadeev, 2015-05-12
@fadeev2010

For Hetzner comparison:
dd if=/dev/zero of=tfl count=512 bs=10240k conv=fdatasync
512+0 records in
512+0 records out
5368709120 bytes (5.4 GB) copied, 11.1591 s, 481 MB/s
Denis Verbin
Addition:
sudo smartctl --all /dev/sda | grep -E 'Power_On_Hours|Power_Cycle_Count'
Power On Hours - the number of hours spent in the on state. Passport time between failures (MTBF) is selected as a threshold value for it. Typically, the MTBF is huge, and it is unlikely that this parameter will reach a critical threshold. But even in this case, the failure of the disk is completely optional.
Power Cycle Count is the number of complete disk on/off cycles. This and the previous attribute can be used to estimate, for example, how much the disk was used before purchase.

D
Dmitry Shinkar, 2015-05-14
@DeadCowsDontMoo

They provide both 100 and 200 , but a little expensive ... - are you sure that it's worth it?)
Read the characteristics of the machines there ...

Didn't find what you were looking for?

Ask your question

Ask a Question

731 491 924 answers to any question