Smith6612

Personally Linksys has been the go-to brand for home routers, and they certainly are a look but as mentioned, they have gone downhill since the days of the WRT54G* line (which are pretty much rock solid). The new UFO and now "curved plate" design units aren't bad, but they do leave something to be desired. 3rd party firmware support and Gigabit ports as standard, for one.

 

Netgear and D-Link have been equally placed at #2 on my list. They make decent units, some firmware quirks of course, but I've never had anyone come to me with issues with them minus a few that have had routers suddenly get up and go into a reboot cycle or have the Wi-Fi chip fail too soon. Lately ASUS has made their way into the market so you may wish to really consider taking a look at them.

 

I would avoid Belkin to be honest, though. Maybe they'll change now that they've purchased Linksys from Cisco, but their routers while they work, have never impressed me with specs or longevity. At least the firmware is based on Linux, which is a plus!

 

Ultimately, get any router with Gigabit ports and you'll be golden. Wireless N/Dual band is a plus.

Sounds good. Keep in mind, I believe Hughes with their Gen4 service took away the unlimited bandwidth usage during the night, and you now have a bonus bucket of bytes each night. So, if you run large downloads during the night, best to let you know about that so you aren't stuck at sub-dial up speeds under the FAP.

If you're out of range, it doesn't mean you cannot get service. Hughes might be able to get you running on their Gen4 satellites, but keep in mind, it might be the same deal with being outside of the usable range of DSL or for that matter, wireless service. It may work, but if the powers that be (usually weather) decide, your service may be far more likely to degrade or go out. With satellite, you're very prone to rain fade if you have a marginal signal as is.

nginx is an amazing piece of software. Light, fast, and as everyone mentions it holds out pretty well against abuse. I use it on my dedicated server to run several websites (mostly hosted for others) that do pick up a bit of traffic running forums. The box itself is often under heavy load, and Apache would start to grind to a halt with the load the machine is under.

The connection limit is for being a good netizen. While there are standards, you can enable pipelining which helps more than spawning a boatload of connections as you go. Just remember, more connections results in more server load and more overhead on your network and the server, so use it in moderation. Some servers are aggressive with the amount of connections they allow as well, and will firewall you if you spawn too many in a short amount of time.

Agreed. For the most part, the only problems I've ever seen with my connection were due to some sort of poor transit between some connection in my ISP's network to some connection who knows where. What does bother me however is, as you've mentioned, the networks were designed so long ago to where you've got many redundant systems tying into a single failure point, or many going into few which causes issues such as slow speeds, or, which Verizon is currently overdue for this year, routing outages in NYC that take out 95% of my Internet connectivity completely but do not destroy speeds. If it's not that, you've also got plenty of older systems that just have not been invested in and are having to handle loads that they can take, but don't allow for smooth failover. Or for that matter, old systems that choke down at night because they were set up with too much oversubscription (such as my local Cable Company's network/backbone).

Then of course, there are providers who just don't plan their networks for double growth (Moore's Law, in perspective) or upgrade them fast enough for whatever reason. I know my second ISP, Frontier Communnications is one of these. Their services are known to slow down at night due to lack of transit capacity or overloaded routers during the night time hours. They're starting to fix all of that up now that they're *FINALLY* bringing out the bigger guns. My Frontier connection before the days of streaming and YouTube in HD was pretty solid. Come all of the big streaming services like Netflix, and higher quality video, I started seeing speeds dropping from 3Mbps down to 1Mbps at night, perhaps lower depending on the day as the edge router at the ISP got overloaded, the old SONET rings not having been upgraded, and so on.

I do know that part of the problem here is with an ISP saying they don't guarantee speeds (and none can), but if it's something that happens routinely according to events, or time of day it's something they have to work to fix up.

Upgrades have been getting done though, which is fortunately something that is being done here in the East. The problem is with old last mile infrustructure and underutilized long haul cable, and of course poor routing to other areas.

Looks like the Datacenter might be having some issues with their New York City to Washington DC transit. I'm getting some awful speeds to the East Coast server tonight. The other two servers, Dallas, TX and Seattle, WA are fine however!

Microsoft Windows [Version 6.0.6002]

Copyright © 2006 Microsoft Corporation. All rights reserved.

C:\Users\>tracert dc.testmy.net

Tracing route to dc.testmy.net [184.173.139.122]

over a maximum of 30 hops:

1 <1 ms <1 ms <1 ms 192.168.3.1

2 8 ms 7 ms 7 ms 10.41.15.1

3 7 ms 6 ms 6 ms P1-0.BFLONY-LCR-01.verizon-gni.net [130.81.195.144]

4 16 ms 16 ms 17 ms so-6-3-0-0.NWRK-BB-RTR1.verizon-gni.net [130.81.28.138]

5 19 ms 18 ms 19 ms 0.xe-8-1-0.XL3.NYC1.ALTER.NET [152.63.5.213]

6 19 ms 19 ms 19 ms 0.xe-2-0-1.XT1.NYC4.ALTER.NET [152.63.20.233]

7 19 ms 20 ms 18 ms GigabitEthernet6-0-0.GW1.NYC4.ALTER.NET [152.63.20.49]

8 19 ms 19 ms 18 ms teliasonera-test.customer.alter.net [157.130.255.206]

9 20 ms 25 ms 19 ms nyk-b6-link.telia.net [213.155.130.33]

10 19 ms 19 ms 35 ms xe-0-0-1.bbr02.tl01.nyc01.networklayer.com [213.248.72.174]

11 31 ms 35 ms 32 ms ae7.bbr01.tl01.nyc01.networklayer.com [173.192.18.176]

12 76 ms 74 ms 72 ms ae1.bbr01.eq01.wdc02.networklayer.com [173.192.18.156]

13 74 ms 76 ms 77 ms ae0.dar01.sr01.wdc01.networklayer.com [173.192.18.197]

14 * 83 ms 76 ms po1.fcr03.sr02.wdc01.networklayer.com [208.43.118.157]

15 78 ms 77 ms 78 ms 184.173.139.122-static.reverse.softlayer.com [184.173.139.122]

Trace complete.

C:\Users\>tracert testmy.net

Tracing route to testmy.net [174.120.187.140]

over a maximum of 30 hops:

1 <1 ms <1 ms <1 ms 192.168.3.1

2 8 ms 7 ms 7 ms 10.41.15.1

3 6 ms 7 ms 6 ms P2-0.BFLONY-LCR-01.verizon-gni.net [130.81.44.232]

4 17 ms 17 ms 17 ms so-6-3-0-0.NWRK-BB-RTR1.verizon-gni.net [130.81.28.138]

5 17 ms 16 ms 16 ms 0.xe-6-0-4.XL3.EWR6.ALTER.NET [152.63.4.77]

6 18 ms 17 ms 17 ms 0.so-1-0-1.XT1.NYC4.ALTER.NET [152.63.0.213]

7 18 ms 17 ms 18 ms GigabitEthernet4-0-0.GW1.NYC4.ALTER.NET [152.63.20.97]

8 19 ms 18 ms 18 ms teliasonera-test.customer.alter.net [157.130.255.206]

9 19 ms 18 ms 18 ms nyk-b6-link.telia.net [213.155.130.33]

10 70 ms 42 ms 18 ms xe-0-0-1.bbr02.tl01.nyc01.networklayer.com [213.248.72.174]

11 44 ms 44 ms 44 ms ae1.bbr01.eq01.chi01.networklayer.com [173.192.18.132]

12 68 ms 67 ms 68 ms ae20.bbr01.eq01.dal03.networklayer.com [173.192.18.136]

13 70 ms 71 ms 69 ms po31.dsr01.dllstx3.networklayer.com [173.192.18.225]

14 66 ms 66 ms 67 ms te2-1.dsr01.dllstx2.networklayer.com [70.87.255.66]

15 68 ms 67 ms 68 ms a.ff.5746.static.theplanet.com [70.87.255.10]

16 67 ms 66 ms 67 ms 8c.bb.78ae.static.theplanet.com [174.120.187.140]

Trace complete.

C:\Users\>tracert west.testmy.net

Tracing route to west.testmy.net [50.23.138.74]

over a maximum of 30 hops:

1 <1 ms <1 ms <1 ms 192.168.3.1

2 7 ms 7 ms 7 ms 10.41.15.1

3 7 ms 6 ms 6 ms P1-0.BFLONY-LCR-01.verizon-gni.net [130.81.195.144]

4 16 ms 16 ms 16 ms so-6-3-0-0.NWRK-BB-RTR1.verizon-gni.net [130.81.28.138]

5 20 ms 20 ms 19 ms 0.xe-8-1-0.XL3.NYC1.ALTER.NET [152.63.5.213]

6 19 ms 20 ms 19 ms 0.xe-2-0-1.XT1.NYC4.ALTER.NET [152.63.20.233]

7 18 ms 19 ms 18 ms GigabitEthernet6-0-0.GW1.NYC4.ALTER.NET [152.63.20.49]

8 19 ms 18 ms 19 ms teliasonera-test.customer.alter.net [157.130.255.206]

9 19 ms 18 ms 19 ms nyk-b6-link.telia.net [213.155.130.33]

10 19 ms 18 ms 19 ms xe-0-0-1.bbr02.tl01.nyc01.networklayer.com [213.248.72.174]

11 46 ms 46 ms 46 ms ae1.bbr01.eq01.chi01.networklayer.com [173.192.18.132]

12 46 ms 45 ms 48 ms ae7.bbr02.eq01.chi01.networklayer.com [173.192.18.171]

13 74 ms 73 ms 74 ms ae1.bbr02.cs01.den01.networklayer.com [173.192.18.131]

14 81 ms 80 ms 80 ms ae7.bbr01.cs01.den01.networklayer.com [173.192.18.168]

15 102 ms 102 ms 102 ms ae0.bbr01.wb01.sea02.networklayer.com [173.192.18.144]

16 100 ms 99 ms 99 ms ae0.dar02.sr01.sea01.networklayer.com [173.192.18.159]

17 100 ms 100 ms 100 ms po2.fcr01.sr01.sea01.networklayer.com [67.228.118.138]

18 102 ms 102 ms 101 ms 50.23.138.74-static.reverse.softlayer.com [50.23.138.74]

Trace complete.

C:\Users\>

Just some food for thought. Wonder if there's anything big going on recently that would have messed with the transit. I know in recent years my ISP has been seeing problems with their network more often than not in regions such as New York and Virginia with poor speeds whereas before problems were often more localized, but they never seem to affect me. I wonder what the deal is. Fibers getting cut, capacity not being upgraded fast enough, poor routing, etc. I also wonder what the deal is with some of the backbones out there that still use older protocols such as ATM. My ISP for example is still using SONET rings middle mile and last mile. Last mile being DSL service which is what I'm on, and middle mile being the transit between POPs and COs until it gets to what is most likely an Ethernet backbone.

Just throwing what's running through my mind out there. Not so much a question but if anyone wants to chime in and just talk about ISP things, that'd be nice

FIOS rules --- I've been jealous since they day they released it... and it's still not in my area

Verizon in other parts of my area is slowly starting to force users to it. I don't have it available to me yet. Verizon was supposed to build it four years ago. They just placed FiOS gear near me a year and a half ago but still need to run the fiber by the homes that gear is supposed to serve! The backbone is already there.

I do like the new look and feel of the new site. Not much to say besides that.

I will put it on the record too, since it's generally not obvious with the way I post. I'm not a Mac hater and I will use a Mac, but I am not always fond of the way Apple does things, the same as other companies go. Microsoft I haven't been a fan of lately due to Windows 8 (which I do have a copy of). Honestly, the only reason I still run Windows here is due to games. If it weren't for Games running poorly in WINE or crashing due to DRM (this stuff makes no sense) hating WINE/PlayOn I would be running Fedora Linux. Many applications I use which are open source run significantly faster on a Linux/Unix system.

Now for that Steam Linux support... I know I can get the Mac version of Steam loaded in but it's a bit buggy.

Ooops Smith6612 sorry I missed your post and links ,

Hey, it's ok! I do the same in many other forums. Half the time I don't thread the full thread before posting something if I'm really tired.

Interesting...

Did you install anything special to the machine at all that you think may be interfering with the time? If it's not due to CMOS battery or power issues on the board, it definitely sounds like something running on the system itself messing with the time if it isn't in constant update with NTP. Sort of what SOHO routers do when you reboot them but not exactly like that.

The thing with Apple and their operating system, and the reason why it doesn't "quite" bog down as much as Windows is known to do is due to the fact that it is practically a highly customized varient of BSD, a flavor of Unix. I like to call it "Bastardized" since you can pretty much run any Unix application you'd like on a Mac out of the box, but there are some things that just don't make sense. Apple did after all, take forever to become POSIX compliant, which began starting with Lion. But otherwise, UNIX as a heart it's obviously going to be a bit more stable. The file system I'm sure also has better management compared to NTFS, but I suppose it also depends on your use since each File System has their own performance characteristics too.

FYI: I have a Windows 3.1 machine that still continues to run from this day, running an Intel 386. I also have a Dell Inspiron from 1999 with one of the original Mobile Pentium III CPUs, that thing continues to run Windows 2000 (and DSL or Knoppix, depending on whichever I decide to choose) and with the original battery, somehow it still holds 1 hour of charge. Some old desktops I built in the 90s and early 2000s are still going as well. All of my hardware goes through quite a bit of use.

On all of the 16 systems I have at home (12 of which are still in full time use, all of the desktops are ones I've built), I've probably only had one GPU in one of the gaming systems go due to a design flaw with the card, which I could replace on the spot. One of the machines from 2002 actually runs Windows 7 Ultimate N 32-bit, and the thing is booting in 25 seconds flat from POST to desktop. Other than that, no hardware problems.

Anyways, as far as the Mac goes, $1.7 Grand is a bit hefty for a machine with only an i5, 13.3" display and possibly Integrated Graphics (Sandybridge). For that sort of price I'd expect dedicated GPU from NVIDIA or AMD, a mid-range i7, perhaps an SSD along with a good chunk of RAM (8GB of DDR3 goodness with a high clock), oh and of course plenty of room to allow the fans to breathe, something Macs have issues with. If it's Intel HD Graphics anything with an I5 at 13.3" you're looking at a $1,000 machine from Apple with a Mechanical HDD and probably 4-6GB of RAM. Macs are well-built machines though, I will say that. I see plenty of older machines continuing to run just as well as they did the day they were unboxed.

Oh yeah, UEFI = awesome. Macs have been in the EFI game for a while, which is amazing and part of the reason why they do boot quickly. BIOS is a dinosaur. UEFI support is still iffy on Mac but once you get it working, any capable OS takes literally no time at all on an SSD. 2 second boot time anyone?

Lol. I do like Google's homepage for the Star Trek memorial.

In celebration of Star Trek's anniversary I was passing around these two vidoes in the office:

On some phones such as the Samsung Galaxy S III and on some Motorola models I've seen with Google Chrome installed on an Android phone running Jellybean an upward of 40+Mbps/28Mbps on LTE here on testmy.net. It wasn't consistent though. Sometimes would be 5Mbps, sometimes it would be at some high speeds.

Here's some more food for thought. Did Comcast give you a DOCSIS 3.0 modem? 30Mbps/6Mbps has almost an indefinite requirement for a DOCSIS 3.0 modem especially due to the upstream and downstream requirements. 30Mbps is near the limit of most QAM265 systems per downstream channel. If they gave you a D2 modem, get it swapped for a D3. That should fix the slow speed issues you're seeing. At the lower packages, a DOCSIS 2.0 modem does work fine.

EDIT: Re-read the orignal post. That's a DOCSIS 2.0 modem. Also, from my own experience with the RCA/Thompson DCM425 units, those things are awful. It may just be the way Time Warner configures them or their firmware though! With my own experience, the RCA modem you have chokes at around 10Mbps, and often locks transfers at an unstable 8-12Mbps, and while it can do PowerBoost up to 30Mbps on 10-15Mbps profiles, it really struggles in real life transfers. I'm not quite sure how it performs in bridge mode though!

QoS being enabled on a router is a good way to overload the CPU on the router itself its usually a good idea to leave it disabled, in most home uses it won't serve a real purpose anyway

~ Mark

True, considering most home routers are already subject to ISP QoS policies at the edge and the home routers already have too much other nonsense running on them to bog them down. Not much you can do when your router's just going to queue up data at the WAN as far as downstream QoS/Traffic Shaping goes. Upstream QoS unless you have FiOS or another symmetrical fiber plan is almost a definite must though. Any router that was built to do QoS in the hardware itself (which is significantly faster and does not tear at the CPU as much) would be suggested for QoS in the home.

My router at home is a few years old, but it has a 532Mhz CPU from Intel that is ARM-based. It's 32-bit, not exactly 64-bit or Dual core like the newer generations of the router I have but the thing is a powerhouse. It's WAN <> LAN Throughput exceeds 100Mbps and that's with QoS services enabled. I don't run anything such as a Samba share or a Print server off of it which cuts down on all the rediculousness, but it is quite a workhorse. CPU load hardly has to work.

Time: 06:58:15 up 284 days, 8:22, load average: 0.00, 0.00, 0.00

I know with the ASUS routers and even D-Links there are these fancy Automatic QoS settings that are sometimes enabled by default. I would definitely try disabling any QoS setting in the router that relates to the WAN connection. Wireless QoS such as WMM can remain on and must be on for Wireless N devices. Also, some newer cable modems are coming with firewalls built-in to them. If your provider uses Ubee or Motorola Gateways for example, I find the firewall in the modem is often to blame for poor speeds, and these often include settings under the guise of "IP Flood Detection" or another term. Shutting that off entirely and bridging the modem too helps.

But definitely, rule out all QoS/GameBooster settings and also rule out any hardware-based firewalls. If you're still seeing issues, definitely check your cable signal levels too. You may be erroring out just enough to kill speeds but not enough to kill the connection entirely. That stuff isn't just with DSL and Fiber

So you're basically looking for more monitors, probably ones with a higher resolution as well correct? If this is for a Mac, what model Mac do you have and what GPU(s) are in it?

My home results:

Nothing fantastic. This is actually a 1Mbps/384kbps line I pay $20/m for but I have it at 5Mbps/768kbps for various reasons.

I also have DSL through Frontier:

This is 3Mbps/384kbps for $42 a month. I've got double the upload however due to various reasons as well.

My home results:

Nothing fantastic. I pay for only 1Mbps/384kbps from Verizon for DSL, but I'm at 5Mbps/768kbps for various reasons . There's also QoS on the Verizon line on the upstream so you're not seeing the full pipe as a result. We use a lot of upload here along with the download and I hate having the latency shoot in the thousands or my download being decreased under heavy uploading. Also, I get 5Mbps for pretty much nothing as I'm the only person in the neighborhood who was given a speed higher than 3Mbps as I'm testing something for the ISP.

I also have DSL through Frontier:

I pay for 3Mbps/384kbps but I receive 3Mbps/768kbps for some various reasons as well . There's no QoS on this connection as Frontier's BRAS is ignoring all QoS flags I set, so the reason I have double the upload I should have is due to the fact that I worked with someone to identify the reason why I couldn't throttle my upstream and keep that line running well. As a trade-off/experiment an upload boost was given, so I do have more upload from them than business customers get on their DSL product

My home results:

Nothing fantastic. I pay for only 1Mbps/384kbps from Verizon for DSL, but I'm at 5Mbps/768kbps for various reasons . There's also QoS on the Verizon line on the upstream so you're not seeing the full pipe as a result. We use a lot of upload here along with the download and I hate having the latency shoot in the thousands or my download being decreased under heavy uploading. Also, I get 5Mbps for pretty much nothing as I'm the only person in the neighborhood who was given a speed higher than 3Mbps as I'm testing something for the ISP.

I also have DSL through Frontier:

I pay for 3Mbps/384kbps but I receive 3Mbps/768kbps for some various reasons as well . There's no QoS on this connection as Frontier's BRAS is ignoring all QoS flags I set, so the reason I have double the upload I should have is due to the fact that I worked with someone to identify the reason why I couldn't throttle my upstream and keep that line running well. As a trade-off/experiment an upload boost was given, so I do have more upload from them than business customers get on their DSL product

Sounds good.

In case you're interested too I do hang around on sites such as DSLReports or the Verizon forums helping others out too, so I am pretty well known amongst DSL-related forums to help resolve connectivity problems. I'll gladly help folks out here when I have my free time as I do on other sites.

You're certainly welcome! I can confirm the test is fixed on my end now. It seems to be uploading the correct amount of data and the expected result is coming back too. Thanks for fixing that bug and also for the promotion! I'm also not seeing any more connection resets to the East coast server, but time will tell I suppose!

I do agree, programming can be a pain at times. I'm not a programmer but I do know HTML, PHP, Lua and C++ a little bit, but for working at a datacenter where you have to deal with an enslaught of programmers wondering why their code doesn't work on the servers properly, I'm sure I've seen quite a bit as it is

I just ran a few more tests. I'm afraid I may still be seeing this problem but it disappears when I press the "Re-Test" button at the top of the page. The tests are showing half of the upload rate i'm actually hitting. A 384KB file completed in around 4 seconds, whereas the test states my speed was around 340kbps which wouldn't upload that big of a file that quick.

I do notice something interesting here and I believe I discovered the issue I'm seeing with upload testing. I noticed that when I performed a test initially, the test would upload more data than it was supposed to. For example, selecting East Coast (closest to me) and then choosing 1.5MB of data would upload 2.7MB of data instead. If I press Re-test at the results page, the correct amount of data is uploaded and I see a proper result. I see this on each server, by the way.

Here's a video I uploaded to YouTube to demonstrate this:

.

Keep an eye on my line monitoring gadget. Notice, it will peg the upload at full speed but the first test always takes longer to run even if it got up to speed right away! The tally of data is showing extra data being uploaded, which I presume is skewing the results. I'm not quite sure what's happening here, but I am using Firefox 15 at the moment. I do NOT see this issue in IE, however. Google Chrome I'm not quite sure about nor Safari.

Makes sense. I'll test out the pages once again when I get home to see if things are any improved. It makes sense the old testmy.net site would be on the main server (Dallas?) as back then there was only one server if I remember correctly to test off of.

Thanks for the warm welcomes by the way