New user's registration have been closed due to high spamming and low trafic on this forum. Please contact forum admins directly if you need an account. Thanks !

Boosting Bubba2 transfer rate

Got problems with your B2 or B3? Share and get helped!
Post Reply
mojs
Posts: 5
Joined: 20 Jul 2010, 13:33

Boosting Bubba2 transfer rate

Post by mojs »

Hi guys, a short summary first.
I managed to get the transfer rate over NFSv3 from 7.65MiB /s to 8.75 MiB/s by removing iptables.

I timed the transfer speed by downloading a file over NFSv3 (default settings).
The copying was done via the following command

Code: Select all

cat <big_file_at_nfs_server> /dev/null
The transfer speed was monitored with bmon.

First I Tested with IP-tables enabled, and set to the following rules (Bubba2 default?):

Code: Select all

Chain INPUT (policy DROP)
target     prot opt source               destination
REJECT     tcp  --  anywhere             anywhere            tcp flags:SYN,ACK/SYN,ACK state NEW reject-with tcp-reset
DROP       tcp  --  anywhere             anywhere            tcp flags:!FIN,SYN,RST,ACK/SYN state NEW
ACCEPT     all  --  anywhere             anywhere            state RELATED,ESTABLISHED
ACCEPT     all  --  anywhere             anywhere
ACCEPT     all  --  anywhere             anywhere
ACCEPT     icmp --  anywhere             anywhere            icmp time-exceeded
ACCEPT     icmp --  anywhere             anywhere            icmp echo-request
ACCEPT     icmp --  anywhere             anywhere            icmp fragmentation-needed
ACCEPT     tcp  --  anywhere             anywhere            tcp dpt:ssh
ACCEPT     tcp  --  anywhere             anywhere            tcp dpt:www
ACCEPT     tcp  --  anywhere             anywhere            tcp dpt:https
ACCEPT     tcp  --  anywhere             anywhere            tcp dpt:imap2
ACCEPT     tcp  --  anywhere             anywhere            tcp dpt:imaps

Chain FORWARD (policy DROP)
target     prot opt source               destination
ACCEPT     all  --  anywhere             anywhere
ACCEPT     all  --  anywhere             anywhere            state RELATED,ESTABLISHED
ACCEPT     icmp --  anywhere             anywhere            icmp fragmentation-needed

Chain OUTPUT (policy ACCEPT)
target     prot opt source               destination
Speed seems to peak at 7.69MiB/s
Eyeballing an average to around 7.65 MiB/s


Now turning of most of the iptables rules,
modifing the /etc/network/firewall.conf file to look like this:

Code: Select all

*filter
:INPUT ACCEPT [0:0]
:FORWARD ACCEPT [0:0]
:OUTPUT ACCEPT [0:0]
COMMIT
*nat
:PREROUTING ACCEPT [0:0]
:POSTROUTING ACCEPT [0:0]
:OUTPUT ACCEPT [0:0]
COMMIT
Then restarting the firewall.

[code|
/etc/init.d/bubba-firewall restart
[/code]

Speed seems to peak at 7.91MiB/s
Eyeballing an average to around 7.82MiB/s
just a minor improvement.

lsmod shows the following:

Code: Select all

Module                  Size  Used by
ipt_REJECT              1992  0
xt_state                 908  0
xt_tcpudp               1836  0
iptable_filter           924  0
ipt_MASQUERADE          1196  0
iptable_nat             3312  0
ip_tables               9984  2 iptable_filter,iptable_nat
x_tables                9716  6 ipt_REJECT,xt_state,xt_tcpudp,ipt_MASQUERADE,iptable_nat,ip_tables
nfsd                   87200  11
nfs_acl                 2176  1 nfsd
exportfs                3044  1 nfsd
ipv6                  228132  32
nf_conntrack_irc        2784  0
nf_nat_ftp              1460  0
nf_nat                 11654  3 ipt_MASQUERADE,iptable_nat,nf_nat_ftp
nf_conntrack_ipv4       8656  3 iptable_nat,nf_nat
nf_defrag_ipv4           756  1 nf_conntrack_ipv4
nf_conntrack_ftp        4892  1 nf_nat_ftp
nf_conntrack           44532  8 xt_state,ipt_MASQUERADE,iptable_nat,nf_conntrack_irc,nf_nat_ftp,nf_nat,nf_conntrack_ipv4,nf_conntrack_ftp
dm_snapshot            20528  0
dm_mirror              12372  0
dm_region_hash          6220  1 dm_mirror
dm_log                  7344  2 dm_mirror,dm_region_hash
mpc8xxx_wdt             3640  0
Let's remove the modules related to iptables.

Code: Select all

rmmod ipt_REJECT
rmmod xt_state
rmmod xt_tcpudp
rmmod iptable_filter
rmmod ipt_MASQUERADE
rmmod iptable_nat
rmmod ip_tables
rmmod x_tables
lsmod does now show the following:

Code: Select all

Module                  Size  Used by
nfsd                   87200  11
nfs_acl                 2176  1 nfsd
exportfs                3044  1 nfsd
ipv6                  228132  32
nf_conntrack_irc        2784  0
nf_nat_ftp              1460  0
nf_nat                 11654  1 nf_nat_ftp
nf_conntrack_ipv4       8656  2 nf_nat
nf_defrag_ipv4           756  1 nf_conntrack_ipv4
nf_conntrack_ftp        4892  1 nf_nat_ftp
nf_conntrack           44532  5 nf_conntrack_irc,nf_nat_ftp,nf_nat,nf_conntrack_ipv4,nf_conntrack_ftp
dm_snapshot            20528  0
dm_mirror              12372  0
dm_region_hash          6220  1 dm_mirror
dm_log                  7344  2 dm_mirror,dm_region_hash
mpc8xxx_wdt             3640  0
Now we have some clear improvments!

Speed seems to peak at 8.80MiB/s
Eyeballing an average to around 8.75 MiB/s

For everyone that is just using the Bubba 2 as a home-server, completely removing the IP-tables should be a viable solution for a nice transfer rate improvement.
For everyone using the Bubba 2 as a router, a different set of IP-filtering software then ip-tables might show some improvements. If someone have the time to try it out, please let me know!
Post Reply