Read speed too bad on NAS share on VM

abhisheks77 · 10-30-2014, 12:08 PM

Hi,
We have two servers in scenario (vmsoldot01 is Oracle VM with Linux and tldtppod15 is physical Linux server). One NAS share is mounted on both servers with similar permissions and access. But READ speed is too bad on virtual in comparison to physical server.
While trying to diagnose this, I see RX-DRP value is high. Can I assume that this is (also) playing role in it ? What else can I check ? Any pointer ?

Quote:

[root@vmsoldot01 ~]# dd if=/dev/zero of=/tmp/bdev/testing/vmsoldot01 bs=1M count=1024 oflag=direct
1024+0 records in
1024+0 records out
1073741824 bytes (1.1 GB) copied, 9.73523 s, 110 MB/s
[root@vmsoldot01 ~]# dd if=/tmp/bdev/testing/vmsoldot01 of=/dev/null bs=1M count=1024 iflag=direct
1024+0 records in
1024+0 records out
1073741824 bytes (1.1 GB) copied, 1246.77 s, 861 kB/s
[root@vmsoldot01 ~]# dd if=/tmp/bdev/testing/vmsoldot01 of=/dev/null iflag=direct
2097152+0 records in
2097152+0 records out
1073741824 bytes (1.1 GB) copied, 471.854 s, 2.3 MB/s
[root@vmsoldot01 ~]#
------------------------------------
[root@tldtppod15 tmp]# dd if=/dev/zero of=/tmp/bdev/testing/tldtppod15 bs=1M count=1024 oflag=direct
1024+0 records in
1024+0 records out
1073741824 bytes (1.1 GB) copied, 100.014 s, 10.7 MB/s
[root@tldtppod15 tmp]# dd if=/tmp/bdev/testing/tldtppod15 of=/dev/null bs=1M count=1024 iflag=direct
1024+0 records in
1024+0 records out
1073741824 bytes (1.1 GB) copied, 92.4556 s, 11.6 MB/s
[root@tldtppod15 tmp]# dd if=/tmp/bdev/testing/tldtppod15 of=/dev/null iflag=direct
2097152+0 records in
2097152+0 records out
1073741824 bytes (1.1 GB) copied, 1112 s, 966 kB/s
--------------------------------------
[root@vmsoldot01 ~]# netstat -i
Kernel Interface table
Iface MTU Met RX-OK RX-ERR RX-DRP RX-OVR TX-OK TX-ERR TX-DRP TX-OVR Flg
eth0 1500 0 22895564 0 101660 0 19052738 0 18 0 BMRU
eth1 1500 0 822638 0 221 0 13997 0 17 0 BMRU
lo 65536 0 4281740 0 0 0 4281740 0 0 0 LRU
[root@vmsoldot01 ~]#
------------------------------------
[root@tldtppod15 ~]# netstat -i
Kernel Interface table
Iface MTU Met RX-OK RX-ERR RX-DRP RX-OVR TX-OK TX-ERR TX-DRP TX-OVR Flg
bond0 1500 0 30489184 0 5095 0 21940113 0 0 0 BMmRU
104b03261a 1500 0 2587924 0 673723 0 0 0 0 0 BMRU
108e4ab487 1500 0 804990 0 7994 0 0 0 0 0 BMRU
100225caed 1500 0 4589169 0 128230 0 0 0 0 0 BMRU
105042eef6 1500 0 10 0 0 0 6 0 0 0 BMRU
1087241aa6 1500 0 115495 0 2590 0 0 0 0 0 BMRU
1061370512 1500 0 2827742 0 21722 0 0 0 0 0 BMRU
bond0.839 1500 0 2837706 0 0 0 0 0 0 0 BMRU
bond0.840 1500 0 4592116 0 0 0 0 0 0 0 BMRU
bond0.842 1500 0 23051901 0 0 0 19030177 0 0 0 BMRU
bond1 1500 0 12194321 0 93820 0 12223822 0 0 0 BMmRU
bond1.350 1500 0 1035850 0 0 0 1068301 0 0 0 BMRU
bond1.941 1500 0 11062114 0 20537 0 9803285 0 0 0 BMRU
bond1.981 1500 0 10 0 0 0 6 0 0 0 BMRU
bond2 1500 0 949087 0 5092 0 309561 0 0 0 BMmRU
bond2.151 1500 0 116429 0 0 0 0 0 0 0 BMRU
bond2.152 1500 0 825020 0 0 0 13997 0 0 0 BMRU
eth4 1500 0 12105592 0 1 0 12223822 0 0 0 BMsRU
eth5 1500 0 949087 0 0 0 309561 0 0 0 BMsRU
eth6 1500 0 88729 0 88730 0 0 0 0 0 BMsRU
eth7 1500 0 30489184 0 0 0 21940113 0 0 0 BMsRU
lo 65536 0 174 0 0 0 174 0 0 0 LRU
vif1.0 1500 0 19030177 0 0 0 22864644 0 0 0 BMRU
vif1.1 1500 0 13997 0 0 0 820349 0 0 0 BMRU
[root@tldtppod15 ~]#

nini09 · 10-31-2014, 02:57 PM

In general, network performance on virtual machine is slow. You can follow some guide to tuning the performance.

jefro · 10-31-2014, 03:53 PM

A vm can get close to or in some rare cases faster than a physical computer. If you do notice huge differences then we have to assume one of a few things. One is physical host support. Two is config as above notes. Three is enough resources added to client. Maybe a few other issues. Get rid of swap on vm maybe, be sure to add in and have plenty of ram for host and client.

abhisheks77 · 10-31-2014, 04:29 PM

Thanks. I am checking more on it. Also checking tuning. Will post, if I find something.