I am considering starting up a passive BWauth in order to understand how they work and fix bugs. What is the minimum and ideal hardware configuration for a BWauth? Have my eye on an OVH config with unmetered 1G, a fuzzy promise of 500mbps minimum bandwidth and a 3.4GHz 4core/8thread AES-NI CPU: http://ark.intel.com/products/80910 The server could be turned up in any of their facilities: https://www.ovh.com/us/about-us/datacenters.xml Or is 10G a requirement? (out of my league) Is all code here https://gitweb.torproject.org/torflow.git/tree/NetworkScanners/BwAuthority or is there more? Are uncommitted fixes lying around anywhere? Thanks
A 10GB network connection is not a requirement, 1GB would be fine, 500MB would also be fine. Mine is 4 core, Intel(R) Xeon(R) CPU E5606 @ 2.13GHz w/ 8GB of RAM. Everything is in torflow, I'm not aware of any other code. -tom On 2 November 2015 at 17:26, <starlight.2015q3@binnacle.cx> wrote:
I am considering starting up a passive BWauth in order to understand how they work and fix bugs.
What is the minimum and ideal hardware configuration for a BWauth?
Have my eye on an OVH config with unmetered 1G, a fuzzy promise of 500mbps minimum bandwidth and a 3.4GHz 4core/8thread AES-NI CPU: http://ark.intel.com/products/80910
The server could be turned up in any of their facilities: https://www.ovh.com/us/about-us/datacenters.xml
Or is 10G a requirement? (out of my league)
Is all code here
https://gitweb.torproject.org/torflow.git/tree/NetworkScanners/BwAuthority
or is there more? Are uncommitted fixes lying around anywhere?
Thanks
_______________________________________________ tor-dev mailing list tor-dev@lists.torproject.org https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-dev
Yep, I've run Mike Perry's code before. It's all in torflow. I was also on a 1 gbits link, but as I recall it wasn't that saturated so you might be able to get away with a 500 mbits. I last ran his scripts back in July, and the logs and other saved data consumed all remaining disk space. I don't know if I missed a cleanup step somewhere or if the issue was fixed since then, but it's something to be aware of. -- Jesse V. On 11/03/2015 07:29 PM, Tom Ritter wrote:
A 10GB network connection is not a requirement, 1GB would be fine, 500MB would also be fine. Mine is 4 core, Intel(R) Xeon(R) CPU E5606 @ 2.13GHz w/ 8GB of RAM.
Everything is in torflow, I'm not aware of any other code.
-tom
On 2 November 2015 at 17:26, <starlight.2015q3@binnacle.cx> wrote:
I am considering starting up a passive BWauth in order to understand how they work and fix bugs.
What is the minimum and ideal hardware configuration for a BWauth?
Have my eye on an OVH config with unmetered 1G, a fuzzy promise of 500mbps minimum bandwidth and a 3.4GHz 4core/8thread AES-NI CPU: http://ark.intel.com/products/80910
The server could be turned up in any of their facilities: https://www.ovh.com/us/about-us/datacenters.xml
Or is 10G a requirement? (out of my league)
Is all code here
https://gitweb.torproject.org/torflow.git/tree/NetworkScanners/BwAuthority
or is there more? Are uncommitted fixes lying around anywhere?
Thanks
_______________________________________________ tor-dev mailing list tor-dev@lists.torproject.org https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-dev
tor-dev mailing list tor-dev@lists.torproject.org https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-dev
On Tue, Nov 03, 2015 at 09:32:26PM -0800, Jesse V wrote:
Yep, I've run Mike Perry's code before. It's all in torflow. I was also on a 1 gbits link, but as I recall it wasn't that saturated so you might be able to get away with a 500 mbits.
I've been running it on a 100mbit link, and it seems to do ok. It doesn't use that much transit total. It's possible that it's hitting the 100mbit limit at times, but hey, my bwauth numbers don't seem any crazier than the numbers from the other bwauths. :) --Roger
Thanks to all for the feedback. That Torflow works from links slower than those of the fastest relays seems to indicate it's measuring relative path resistance as much or more than absolute bandwidth. I often see it produce sensible results and hope that some tuning and fixing might produce better consistency and faster scans. Probably can't start until January, but I intend to proceed.
participants (4)
-
Jesse V -
Roger Dingledine -
starlight.2015q3＠binnacle.cx -
Tom Ritter