[ Date Index ] [ Thread Index ] [ <= Previous by date / thread ] [ Next by date / thread => ]
|
>> have you tuned your kernel? Well, pfSense is a fairly locked down BSD-based distro – so my answer is “no”, and a longer answer is “er, I’m not sure if I can” Of course – that could also apply to the NAS & Myth boxes. The NAS is based on Debian and has its own installer – so it should be straight forward to find out
what optimisations have been done there…. The Myth box is a complete self-install using Arch with no optimisations what so ever… so there’s an opportunity there – inc getting the NIC up to 1Gb. So, thanks – I’d not considered kernel tuning. Which, I guess is the point of my original post… how would I measure the “tuning”, to know that it’s made any (negative?) affect? Cables are a mix of Cat5e and Cat6 in an accurate representation of my intestine combined with my MP3 player headphone leads... and yep, some run past and are
probably wrapped around a domestic 50Hz white-noise signal injector :o) So, how would I best measure the before & after ? FYI – I’ve read that cat6 best practices are NOT to run cables in long straight lines – in fact, a random mess is more beneficial due to cross-talk being “easier”
at higher frequencies (Cat6 cables are effectively a wave guide rather than an electrical conductor) – ok, we still don’t want 50Hz AC next to them, but you don’t want those neat, cable-tied layouts that we’ve perfected with Cat5e over the years… From: list [mailto:list-bounces@xxxxxxxxxxxxx]
On Behalf Of Joseph Bennie On 15 Sep 2014, at 17:04, Tremayne, Steve <steven.tremayne@xxxxxxxxxxxxx> wrote:
I should’ve mentioned that the “firewall” is another PC (a repurposed thin-client) running pfSense with 1Gb NICs, ticking over quite nicely (according to it’s
RRD graphs) You’ll be surprise how much it slows it down, the nic speed has nothing to do with it, its the latency thats added to do the work on the packet. have you tuned your kernel?
However…. You’ve made me think (a good thing…) I’ve got a 1Gb Netgear “Pro” switch in there… yet the MythTV box is only running at 100Mb…. Hmmm…. hmm - cables’ good point - i recently swapped out some unshielded CAT5e ones for some shielded CAT6 and the difference was v noticeable, you might have interference if running parallel to power cables.
OK, so there’s one issue… J Of course, when I get that set to 1Gb, I’ll be asking the same question J… From: list
[mailto:list-bounces@xxxxxxxxxxxxx] On Behalf Of Joseph Bennie On 15 Sep 2014, at 14:58, Tremayne, Steve <steven.tremayne@xxxxxxxxxxxxx> wrote:
Afternoon… I’m currently transferring some files between my home built NAS (OpenMediaVault – basically Debian with a webserver running on a mdadm RAID 1) and my media PC
(MythTV… based on Arch, using LVM (no RAID)) – all rotating drives. Transferring via rsync seems to be getting about ~3MB/s (~30Mb/sec)… and for 1TB of media, that’ll take a while… which is fine as it’ll run over night. However, I’m now curious as to what’s causing the bottleneck… the network (100Mb cabled through a firewall), the disks (SATAs), rsync over ssh, phase of the moon,
etc… The most probable reason is the firewall ( common home /small business level devices have a throughputs of between 7Mbs and 40Mbs when IPS , IDS are enabled) + collisions on the network (can half the effective speed or worse)
So, I’m just asking as to what tools I can use to test the performance of each section… Quick duckduckgo searches show that hdparm has the ability to do some tests… and iperf seems good for networking… Suggestions? Thanks, Steve -- -- |
-- The Mailing List for the Devon & Cornwall LUG http://mailman.dclug.org.uk/listinfo/list FAQ: http://www.dcglug.org.uk/listfaq