summary: For the person who was asking about the "greatest thing since sliced bread" reviews, the tcp optimizer was that good for Windows XP and earlier. For Vista and later that have auto-tuning stacks.. my guess is that it's mainly the placebo effect. Enabling selective acks & path mtu discovery if not already enabled will help but will not make a dramatic difference like changing the tcp window size did on WinXP.
On the other hand, every little bit helps, so it could be worth a shot. Just make sure you know how to undo whatever it changes (which is why I prefer using regedit. I know what I changed/added and what the old values were) remainder of my comments inline On 2/2/15, Thee Chicago Wolf (MVP) <"."> wrote: > On Sun, 1 Feb 2015 18:25:36 -0500, Lee <[email protected]> wrote: > >>On 2/1/15, Thee Chicago Wolf (MVP) <"."> wrote: >>> On Sat, 31 Jan 2015 10:33:24 -0500, Lee <[email protected]> wrote: >>> >>>>On 1/30/15, WaltS48 <[email protected]> wrote: >>>>> On 01/30/2015 05:39 PM, Richard Owlett wrote: >>>>>> Thee Chicago Wolf [MVP] wrote: >>>>>>> [snip] >>>>>>> Yes, we know there's still a population out there on dial-up. You >>>>>>> should also use TCPOptimizer 3.0.8 (it's Free) to squeeze out a >>>>>>> little >>>>>>> bit more oomph from that old dial-up connection. Used to work well >>>>>>> for >>>>>>> me back in the day. >>>>>> >>>>>> A link to author's site please. >>>>>> A Google search turns up multiple "greatest thing since sliced bread" >>>>>> reviews. >>>>>> What say author(s)? >>>>>> TIA >>>>> >>>>> >>>>> "The program can aid both the novice and the advanced user in tweaking >>>>> related TCP/IP parameters in Windows, making it easy to tune your >>>>> system >>>>> to >>>>> the type of Internet connection used." >>>>> >>>>> <http://www.speedguide.net/downloads.php> >>>> >>>>I looked at it a long time ago -- back when WinXP was all shiny & new. >>>>It was useful if you had a broadband connection; the default 16KB tcp >>>>window size wasn't large enough for even a 10Mb connection & most >>>>people would rather use a GUI than regedit to make registry changes. >>>> >>>>But for dialup, a 16KB tcp window is plenty enough. the way tcp works, the best thruput you can get is (tcp window size) / (round trip time) So let's say we have a 10Mb/s download speed, a 16KB tcp window and the round trip time from pc to server is 20ms. So we've got 16 * 1024 = a tcp window size of 16384 bytes. divide that by .02 & I've got a max download speed of 819,200 bytes/sec or a bit over 6.5Mb/s. For a server further away, like www.mozilla.org that gives me a ping time of about 85ms, I'd be getting less than 1.6 Mb/s thruput on my 10Mb/s link So it's easy to see why a tcp optimizer program would get such glowing reviews for broadband users... run the program, reboot & shazzam! You're finally getting download speeds approximating the download speed you're paying for. But dialup tops out at 56Kb/s on a pristine line. So even with an outrageous round trip time of 1 second the limiting factor is still the dialup line speed, not the window size. >>>> And starting with >>>>Vista, MS made the tcp stack auto-tuning so there are no big gains to >>>>be had using one of those tcp optimizer programs on a currently >>>>supported Windows OS. (in other words, if the OP still has XP it >>>>might help on the library wireless connection) >>>> >>>>So don't expect much from optimizing a dial-up connection, but it >>>>could help a bit if selective acks or path mtu discovery aren't >>>>already enabled. >>>> >>>>Regards, >>>>Lee >>> >>> I still use it on all my WIn 7 boxes, productions workstations (600+) >>> and on my servers. It helps me IMMENSELY, esp. on my printer servers >>> which get slammed with GIGs of print data per hour. Gigs per hour on current hardware is nothing remarkable. The backup staff @work doesn't backup everything over the san yet, so there's still a few backup servers with 10Gb NICs on the net that handle GIGs of backup data per second. >>Ok - I'll bite :) What settings do you use, what changes does tcp >>optimizer then make & what's the before & after response >>time,thruput, or [what?] difference on your print servers? >> >>> There are as many critics of it as there are advocates but the critics >>> are the idiots who make arguments like "it doesn't make my Internet >>> faster." Yeah, no sh*t Sherlock. It won't make a 10Mb connection >>> suddenly go to 20Mbit which s what most people *think* it is supposed >>> to do. That is not its purpose. It's purpose is to make your downloads run closer to line speed by changing the settings that artificially restrict download speed. >>> TCPOptimizer's whole purpose is to take XP through to Win 7's TCP >>> features that are disable by default and properly tune them or enabled >>> them for maximum responsiveness. >> >>Is English your native language? "maximum responsiveness" is an >>extremely vague term, so I'm wondering exactly how one measures it. >>Again, if you have them, the numbers for responsiveness before & after >>using tcp optimizer as well as the settings you use for tcp optimizer >>would be appreciated. >> >>> XP through WIn 7 have ok TCP stacks, Win 7's being decent, but they >>> are, by default, set to the *worst* case scenario by Microsoft >>> themselves. TCPOptimizer just sets things for the *BEST* case >>> scenario...with the worst case scenario in mind (dirty lines, noise, >>> lag, bad TTL, latency, etc.). >>> >>> When used correctly with a competent understanding of networking, >> >>& here I thought I had a competent understanding of networking. Which >>is why I doubt it'll do much for a Win 7 box. >> >>> ... it >>> works and works well. I've always seen my throughput and >>> responsiveness jump up when using it. It's never failed me. >> >>Please enlighten me. What changes need to be made on Win7 to improve >>thruput & responsiveness? >> >>Regards, >>Lee > > Doubting Thomas?. I'll bite back. I do suggest you do some Google > search on anyone who has don't before and after metrics and simple > speedtest.net tests. There are quite a few out there. I was asking you. I've done my own testing & don't care what random people of unknown abilities have to say (which is pretty much what you get with web searches). > For one, the "optimal" setting works well. Two, don't accept the > default TTL suggested. I've found in most cases a TTL of 64 degrades > performance so I always manually delete it and let Windows manage > it.That's the one thing I wish it did not suggest as it is usually two > low and not a standard TTL on any networks I've used it on. Most times > I see TTL between 128 and 256 so a setting of 64 isn't optimal in all > cases. If changing the TTL makes _any_ difference other than the destination becomes reachable or unreachable your network is seriously broken. Learn how traceroute works -- and please don't say you already know; if you did you wouldn't have said a TTL of 64 degrades performance. > As for metrics, I tested with monstrous PDFs and PPTs loaded with 10+ > megabit embedded images that, when spooled, created a 1GB+ spool files > (read: print job, for the rest of us). In my case, sending jobs up to > the print server spooled faster. Since that's the meat and potatoes in > my prod environment, that's where I see the most benefit for my users. > YMMV. > > Beside that, on my personal workstations, networking is just snappier > and more responsive. That's it really it in a nutshell. In other words, you made some changes expecting it would make things better, and now things do seem to be better. Could be the placebo effect, could be things actually did get better.. > If you should > so choose to go and read up on all the settings that it modifies, > you'll learn what they do and why tuning/enabling them can benefit > you. If you're going to troll without doing your homework... But I was doing my homework - asking you questions so I could judge your level of ability :) I would suggest that you need to do your homework -- and the IETF has just made it a lot easier for you w/ a brand new RFC: http://www.rfc-editor.org/rfc/rfc7414.txt A Roadmap for Transmission Control Protocol (TCP) Specification Documents Regards, Lee > > As I said, if you're digging deeper than that, you're falling in with > the "It didn't make my 10Mb connection go to 20Mb so it sucks and > doesn't work." crowd. And I again respond, that's not what it's > purpose is. Thanks for asking. > _______________________________________________ > support-seamonkey mailing list > [email protected] > https://lists.mozilla.org/listinfo/support-seamonkey > _______________________________________________ support-seamonkey mailing list [email protected] https://lists.mozilla.org/listinfo/support-seamonkey
