Hello All,
At the wireless mesh setup in HBCSE, the download speed was very slow, max. 1 to 1.2 Mbps for a single user. How can this speed be optimized so that when we setup meshes all over Mumbai, we get decent speeds with multiple users.
--- Rony wrote:
Hello All, At the wireless mesh setup in HBCSE, the download speed was very slow, max. 1 to 1.2 Mbps for a single user. How can this speed be optimized so that when we setup meshes all over Mumbai, we get decent speeds with multiple users.
I was recollecting the day we were carrying out these experiments of downloading distros (CD ISOs). I remember, you started downloading it on your laptop (with Sabayon) when the signal strength was quite good.
When I started, the signal strength was good, but not better than yours. (indicated by Windows XP's network monitor tool). As soon as I started the download, our speeds were almost divided into two halves, and I got nearly 700 Kbits/ sec.
As suggested by Nagarjuna Sir, we need to carry out benchmark reports and suggest / find out improvements.
Secondly, since this is a wireless mesh, a torrent client may prove a better client for downloading stuff instead of browsers. A better distributed system.
-- FSF of India Associate Fellow - http://www.gnu.org.in S K Somaiya College of ASC- http://www.somaiya.edu/sksasc ubunturos @ freenode
Download prohibited? No problem! To chat from any browser without download, Click Here: http://in.messenger.yahoo.com/webmessengerpromo.php
--- Rony wrote:
Hello All, At the wireless mesh setup in HBCSE, the download speed was very slow, max. 1 to 1.2 Mbps for a single user. How can this speed be optimized so that when we setup meshes all over Mumbai, we get decent speeds with multiple users.
I was recollecting the day we were carrying out these experiments of downloading distros (CD ISOs). I remember, you started downloading it on your laptop (with Sabayon) when the signal strength was quite good.
When I started, the signal strength was good, but not better than yours. (indicated by Windows XP's network monitor tool). As soon as I started the download, our speeds were almost divided into two halves, and I got nearly 700 Kbits/ sec.
As suggested by Nagarjuna Sir, we need to carry out benchmark reports and suggest / find out improvements.
Secondly, since this is a wireless mesh, a torrent client may prove a better client for downloading stuff instead of browsers. A better distributed system.
-- FSF of India Associate Fellow - http://www.gnu.org.in S K Somaiya College of ASC- http://www.somaiya.edu/sksasc ubunturos @ freenode
Download prohibited? No problem! To chat from any browser without download, Click Here: http://in.messenger.yahoo.com/webmessengerpromo.php
On Thu, 2007-06-07 at 20:44 +0530, Rony wrote:
At the wireless mesh setup in HBCSE, the download speed was very slow, max. 1 to 1.2 Mbps for a single user. How can this speed be optimized so that when we setup meshes all over Mumbai, we get decent speeds with multiple users.
With todays technology i.e. wifi ( 54Mbps 802.11g ) one can realistically expect to get 24-30Mbps for p2p connection. So for a mesh network 1-1.2Mbps is pretty good. I think there can be optimizations but I'm not sure.
Anyway, a wireless network cannot be expected to deliver vast quantities of bandwidth with today's tech.
--- Dinesh Joshi wrote:
So for a mesh network 1-1.2Mbps is pretty good.
But would this be the same, when we scale?
Anyway, a wireless network cannot be expected to deliver vast quantities of bandwidth with today's tech.
Actually, it is a wireless *mesh* network. A wired mesh network would be fairly /'messed'/ but a wireless mesh doesn't appear to be, but it should offer *similar* capabilities in terms of performance.
Also, a day would have to be fixed up such that people from the list can bring in their laptops and tests could be conducted on the network for improving it.
-- FSF of India Associate Fellow - http://www.gnu.org.in S K Somaiya College of ASC- http://www.somaiya.edu/sksasc ubunturos @ freenode
__________________________________________________________ Yahoo! India Answers: Share what you know. Learn something new http://in.answers.yahoo.com/
On 6/10/07, Roshan wrote:
So for a mesh network 1-1.2Mbps is pretty good.
But would this be the same, when we scale?
I'm not sure. I'm no expert at this but I can tell you one thing. Mesh networks are _not_ meant for speed. They're meant for high fault tolerance ( due to their inherent redundant nature ).
For a mesh network to be really successful, we'll need to see if there are better ( optimized? ) protocols for these networks. Protocols have a good impact on the throughput.
Actually, it is a wireless *mesh* network. A wired mesh network would be fairly /'messed'/ but a wireless mesh doesn't appear to be, but it should offer *similar* capabilities in terms of performance.
How is wireless mesh network less messier than a wired mesh network?
No. Wireless networks at the moment /cannot/ compete with wired networks on the performance / throughput front. Why? If you study how wired and wireless networks behave you'll understand. Basic problem of wireless networks being that the signals travel through an unbounded medium, signal losses are tremendous, collisions, limited spectrum, so on and so forth...
-
Dinesh Joshi -
Rony -
Roshan