Facebook going slow? - Speed it up with Plusnet.

[#1]

Recently we held a vote on our Community Site asking you which websites you visit most often. More than a quarter of you said that Facebook was the site you visited most regularly. We're happy to tell you that we now have a peering arrangement with Facebook through LINX. This will mean the pages load quicker and you'll be able to see all those application requests even faster. If you're like me you'll reject most of them anyway You can even say hello to us on our Facebook page. So go forth, upload pictures, play Scrabulous, catch up with old friends or just arrange your next night out through Facebook. Chris.

[#2]

Here's a traceroute from before we made the changes. You can see the traffic goes via Telia and that there's some packet loss at the NTT US hops: 1. 126.structure.internal.plus.net                                   0.0% 1137    0.5   0.5   0.4 23.6   1.1 2. fa0-7-6.pih-cr01.plus.net                                         0.0% 1137    0.5   0.4   0.4 13.4   0.8    212.56.71.253    84.93.216.84    fa0-7-149.pih-cr01.plus.net 3. se1-0.ptc-gw1.plus.net                                            0.0% 1137    8.0 15.1   7.8 252.3 30.5 4. ge0-1-0-31.ptn-gw2.plus.net                                       0.1% 1137    9.9 11.5   8.3 97.3   9.4 5. te2-2.pte-gw2.plus.net                                            0.0% 1137    9.2 17.5   7.3 214.0 30.9 6. 80.239.193.141                                                    0.0% 1137    9.3   9.7   7.2 103.1   8.5 7. ldn-bb1-link.telia.net                                            0.2% 1137    8.5 10.0   7.3 105.5   8.6 8. ldn-b4-link.telia.net                                             0.1% 1137    9.4 12.4   8.4 105.9   9.7 9. verio-112851-ldn-b4.telia.net                                     0.0% 1137    9.3 11.5   8.5 107.4   9.3 10. ae-0.r22.londen03.uk.bb.gin.ntt.net                               0.0% 1136   15.0 11.9   7.7 104.8 10.1 11. as-0.r20.nycmny01.us.bb.gin.ntt.net                               0.2% 1136   79.2 83.4 77.3 213.4 11.0 12. ae-0.r21.nycmny01.us.bb.gin.ntt.net                               0.0% 1136   79.2 84.3 78.8 181.1   9.3 13. as-0.r21.asbnva01.us.bb.gin.ntt.net                               0.6% 1136 102.8 101.7 92.3 187.7 10.2    p64-2-0-0.r20.chcgil09.us.bb.gin.ntt.net 14. ae-0.r20.asbnva01.us.bb.gin.ntt.net                               1.1% 1136 109.1 96.5 82.7 183.2 13.5    ae-0.r21.chcgil09.us.bb.gin.ntt.net 15. as-0.r20.snjsca04.us.bb.gin.ntt.net                               0.2% 1136 169.0 164.4 146.3 245.3 10.3    p64-3-1-0.r20.snjsca04.us.bb.gin.ntt.net 16. ae-1.r21.plalca01.us.bb.gin.ntt.net                               0.2% 1136 158.6 163.1 148.7 248.8   9.7 17. xe-4-1.r04.plalca01.us.bb.gin.ntt.net                             0.1% 1136 358.2 199.4 151.4 781.2 71.8 18. xe-4-3.r04.plalca01.us.ce.gin.ntt.net                             0.1% 1136 165.7 166.6 151.1 261.4 10.0 19. xe-4-0-0.br02.sf2p.tfbnw.net                                      0.0% 1136 167.0 170.0 151.4 321.7 17.9 20. te-6-4.csw01a.sf2p.tfbnw.net                                      0.0% 1136 164.1 171.5 151.7 270.1 10.5 21. v501.bsw.sctm.tfbnw.net                                           0.0% 1136 176.6 173.3 153.2 261.9 10.6 Now you can see that the route is shorter and there's no packet loss: 1. 126.structure.internal.plus.net                                   0.0%   119    0.6   0.7   0.4   4.9   0.7 2. fa0-7-6.pih-cr01.plus.net                                         0.0%   119    4.5   0.5   0.4   6.1   0.7    fa0-7-149.pih-cr01.plus.net    212.56.71.253 3. se1-0.ptc-gw1.plus.net                                            0.0%   119   13.8 12.3   7.8 153.9 19.7 4. ge0-1-0-31.ptn-gw2.plus.net                                       0.0%   119   23.3 11.5   8.4 61.6   6.3 5. te2-2.pte-gw2.plus.net                                            0.0%   119   13.8 17.7   7.6 237.2 36.0 6. linx.br01.lon1.tfbnw.net                                          0.0%   119   15.4   9.7   7.6 26.7   2.9 7. xe-2-0-0.br01.ams1.tfbnw.net                                      0.0%   119   17.4 16.9 15.1 30.1   2.3 8. xe-x-x-x.br02.ash1.tfbnw.net                                      0.0%   119 125.6 116.4 109.2 211.9 16.6 9. ge-2-0-0.br02.sf2p.tfbnw.net                                      0.0%   119 193.8 187.3 184.0 281.6 10.1 10. te2-0.csw01b.sf2p.tfbnw.net                                       0.0%   118 198.1 190.2 184.4 207.5   4.9 11. v512.bsw02.sctm.tfbnw.net                                         0.0%   118 196.1 194.6 186.1 310.2 11.9 12. v501.bsw.sctm.tfbnw.net                                           0.0%   118

[#3]

Wow, if only I knew what that meant 🙂

[#4]

lol, basically look at the percentage columns. This shows the amount of data that is 'lost' along the way and needs to be resent. Each horizontal line is a 'stepping stone' or 'hop' that your Internet traffic has to pass through to get to facebook.com. If you look at the second set of results... Less hops = shorter route to Facebook = quicker. Fewer lost packets = less data that has to be resent = quicker. Do you actually use Facebook? If so then it would be interesting to know whether or not you notice an increase in performance?

[#5]

Hi Bob, No don't use Facebook, too old for that, although I believe my son does. I did actually look at your first post again after my one and could see less connections so in theory less things to go wrong, the rest of it, well I guess I'd need to be in your job. 🙂

[#6]

Your first route isn't showing any packet loss. The device at hop 21 has replied to all 1136 packets you have sent it. Any packet loss in the middle is just some routers being a bit overworked/lazy, and not sending all ICMP time outs. That quite often happens, and if you will be hard pressed to find a trace route that this doesn't occur on if you trace it long enough. Additionally your second trace route has a higher RTT time of 194.6ms average, as opposed to 164.4ms average on your first one. Thats 30ms slower to respond! Your second route does have far less hops, although based on what those trace routes is actually showing us, is 'worse' for a person browsing facebook. Less hops do however mean less places for things to go wrong, and those trace routes only show a snapshot of what was happening in a very small space of time.

[#7]

Hmm, I'd put the faster facebook down to the new Firefox version installed tonight. Not been on facebook for a while & it was lots faster to load - some occasional initial page start delays (felt like facebook not net delays) but then fast loads once they started. Rather unscientific, but it did feel faster tonight. Thank you (even if the stats don't add up properly, it seems to deliver. We'll have to see long-term)

[#8]

Used your link for Facebook and had a message pop up from them telling me I was searching too fast and that I must significantly slow down or be banned!

[#9]

so slow why?

[#10]

Thanks for all your feedback guys, it seems to be going well.