The Speed Test

How latency affects networks.

The essential difference is “latency”, or the time taken for a packet to traverse the network. Latency is a major factor in IP performance, and reducing or mitigating the effects of latency is one of the major aspects of network performance engineering.

In order to understand why this is the case it is appropriate to start with the IP protocol itself. IP is a ‘datagram’ protocol. For data to be transferred across an IP network the data is first segmented into packets. Prepended to each packet is an IP protocol header. IP itself makes remarkably few assumptions about the characteristics of the underlying transmission system. IP packets can be discarded, reordered or fragmented into a number of smaller IP packets and remain within the scope of the protocol (ie. the protocol does not assume any particular network service quality, bandwidth, reliability, jitter or error rate).

Layered above IP are two transport protocols, UDP and TCP. UDP is a lightweight extension to IP, implementing an unreliable packet transport protocol. UDP is used as the transport platform real time applications. UDP traffic typically forms some 8% of the total volume of traffic on a public Internet network, and the majority of this traffic is attributable to Domain Name queries - real time traffic, such as voice and video, is still a relatively small proportion of total traffic volumes. UDP traffic is non-adaptive, so that a UDP protocol stack will not attempt to adjust its sending rate to fit within available network capacity.

Latency most effects TCP. TCP is a reliable data transfer protocol. The requirements for reliability in the data transfer implies that the protocol will detect any form of data corruption on the part of the network and retransmit until the data is transferred successfully. This ’stop and retransmit’ implies that there is no fixed rate for data transfer, nor will any implicit timing of packets be preserved by TCP. TCP is not a real-time protocol. TCP attempts to maximise its data transfer rate through dynamic rate adjustment. The way TCP achieves this is to continually test the network to see if a higher data transfer rate can be supported. When TCP encounters packet loss, it assumes that the loss is due to network congestion, and the protocol immediately reduces its data transfer rate.

TCP uses a ’sliding window control’ to send data. By this, it is meant that the sender sends a sequence of packets (a ‘window’), and then holds a copy of these packets while awaiting an acknowledgement (ACK) from the receiver that the packets have arrived. Each time an ACK for new data is received, the window is advanced by one packet, allowing the sender to send the next packet into the network.

The way in which TCP adjusts its rate is by increasing the window each time an ACK for new data is received, and reducing the window size when the sender believes that a packet has been discarded. Typically, in a steady state, the sender sends a burst of packets, and then waits for a corresponding burst of ACKs before sending any further data. When starting a TCP session a control method called “slow start” is used, where the window is increased by one packet each time an ACK is received. Across high latency paths, such as satellite paths, this function can result in packet bursts, where the sender sends a string of packets and then sits idle, waiting for the corresponding string of ACK packets. With TCP, these packet bursts are injected into the network at twice the rate that is available on the data path. The network must perform rate adaptation by using queues to adapt the sending rate to the bottleneck rate. The larger the RTT, or, in other words the greater the latency, the greater this burstiness of the data.

A sender cannot continuously increase its sending rate without limit. At some stage the receiver will signal that its receiving buffer is saturated, or the sender will exhaust its sending buffer, or a network queue resource will become saturated. In the last case this network queue saturation will result in packet loss. When TCP experiences packet loss the TCP sender will immediately halve its sending rate and then enter “congestion avoidance” mode. In congestion avoidance mode the TCP sender will increase its sending rate by one packet every RTT interval. Paradoxically, this is somewhat slower than the “slow start” rate.

For TCP, the critical network characteristic is the latency. The longer the latency, the more insensitive TCP becomes in its efforts to adapt to the network state. As the latency increases, TCP’s rate increase becomes slower, and the traffic pattern becomes more bursty in nature. These two factors combine to reduce the efficiency of the protocol and hence the efficiency of the network to carry data. This leads to the observation that, from a performance perspective and from a network efficiency perspective, it is always a desirable objective to reduce network latency.

If you cannot reduce latency, what can you do? The basic answer is: ‘use very large buffers’. Tuning a TCP stack to support window scaling, selective acknowledgements and large TCP buffers is a very effective means of limiting the worst effects of latency. Setting up large buffers on the routers that connect to the satellite system also has an positive effect on TCP. You cannot always avoid latency, but you can reduce its most harmful effects.

This looks like a cool gadget.  I wouldnt want to spend $400.00 on it though.

 http://www.nytimes.com/2007/09/06/technology/circuits/06pogue.html?bl&ex=1189310400&en=38ac6fe117cc4348&ei=5087%0A

Multimode
Multimode fiber optic cable has a large-diameter core that is much larger than the wavelength of light transmitted, and therefore has multiple pathways of light-several wavelengths of light are used in the fiber core.

Multimode fiber optic cable can be used for most general fiber applications. Use multimode fiber for bringing fiber to the desktop, for adding segments to your existing network, or in smaller applications such as alarm systems. Multimode cable comes with two different core sizes: 50 micron or 62.5 micron.

50- vs. 62.5-micron cable. Although 50-micron fiber features a smaller core, which is the light-carrying portion of the fiber, both 62.5- and 50-micron cable feature the same glass cladding diameter of 125 microns. You can use both in the same types of networks, although 50-micron cable is recommended for premise applications: backbone, horizontal, and intrabuilding connections, and should be considered especially for any new construction and installations. Both types can use either LED or laser light sources.

The main difference between 50-micron and 62.5-micron cable is in bandwidth-50-micron cable features three times the bandwidth of standard 62.5-micron cable, particularly at 850 nm. The 850-nm wavelength is becoming more important as lasers are being used more frequently as a light source.

Other differences are distance and speed. 50-micron cable provides longer link lengths and/or higher speeds in the 850-nm wavelength.
Singlemode
Singlemode fiber optic cable has a small core and only one pathway of light. With only a single wavelength of light passing through its core, singlemode realigns the light toward the center of the core instead of simply bouncing it off the edge of the core as with multimode.

Singlemode is typically used in long-haul network connections spread out over extended areas–longer than a few miles. For example, telcos use it for connections between switching offices. Singlemode cable features a 9-micron glass core.

Duplex
Duplex cable consists of two fibers, usually in a zipcord (side-by-side) style. Use duplex multimode or singlemode fiber optic cable for applications that require simultaneous, bi-directional data transfer. Workstations, fiber switches and servers, fiber modems, and similar hardware require duplex cable. Duplex fiber is available in singlemode and multimode.

Simplex
Simplex fiber optic cable consists of a single fiber, and is used in applications that only require one-way data transfer. For instance, an interstate trucking scale that sends the weight of the truck to a monitoring station or an oil line monitor that sends data about oil flow to a central location. Simplex fiber is available in singlemode and multimode.

Connecting your Sprint Moto Q  to your laptop to use as a modem is a pretty simple task. Just follow these quick instructions:

1) Download and install the software from here. (For XP only)  Use Sprint.com to find for other OS’s.

2) On the Phone, go to start -> Accessories –> Modem Link.

3) click the Activate button.

4) On your laptop, open the Sprint Application.  Then click connect.

That is all there is too it.

With its design, the Motorla Q has earned it the “BlackBerry Killer” nickname. It has the three “S” of the perfect gadget: Slim, Sexy, and Smart. Text can be entered at full speed, thanks to the QWERTY keyboard, and displayed on a nice QVGA display. The inside is nice too: the phone is a 3G CDMA handset that can connect to Sprint’s high-speed EV-DO network. Internet browsing, email and messaging can all be used at reasonable speeds when compared with non-3G phones. Is the Motorola Q the new “King of the hill”? Read this review to get our opinion.

Although it has also been nicknamed RAZRBerry, the Motorola Q does not really go head to head with the BlackBerry line of device because the software and the wireless service are so different. For a personal use, I’d rather compare the Q with the Treo 700w or other QWERTY equipped smart phones.

Design
The beauty of the Motorola Q is pretty much unanimous. Everyone who saw it praised how slim (it’s half of the Treo 700w thickness) and light it is. It is however slightly wider, and although the photos show that it is only slightly wider, the difference is noticeable when holding it in the hand. The overall quality is good but the device feels more like “plastic” and a bit cheaper than the Palm device. That’s also probably why it’s lighter, so I won’t complain.

There’s a mini SD slot for additional storage. Obviously, I wished that hey had opted for a regular SD card, but if this contributes to the smaller form-factor, I’ll just shell $25 and get a new memory card.

I have to congratulate Motorola for using a standard mini USB port to recharge the device. Any portable device should be USB rechargeable. I’m really sick of proprietary charging connectors and it’s funny that the slimmest smart phone is the one to have a standard port. There’s no excuse for not doing it now.

In the back of the unit there’s a digital camera. It can snap 1.3 Megapixel pictures and has a LED flash for dark environment. LED lights (Light Emitting Diode) are great because they are bright but consume only little power. That’s about the best options for cellphones, but don’t expect it to replace your average digital camera, especially in low-light.

The Motorola Q is the only windows mobile device equipped with a jog that comes to mind. Early Sony cellphones and BlackBerry conveniently had one and I think that it’s a good idea to have one on the Q.

Display
The QVGA display is very nice and bright when used indoors. Pretty much anyone who handled the Moto Q praised the screen quality. In direct sunlight, however, the readability comes down by a lot. That’s unfortunate because it’s really sunny in California but to be fair, the smart phone competition isn’t performing any better. Curiously, my Sony Ericsson T610’s display beats them all. Go figure.

The Screen is not a touch screen, which is a little disturbing at first if you’re used to have poke the display with your finger. After a couple of days, I just got used to it, and it became non-issue. During our tests, we found that we used mainly the phone, text messaging and web browsing applications, none of which requires the touch screen.

With a 320 pixels wide screen, web page rendering is better then the Treo 700w. The characters are very readable and users can select different font sizes in Pocket Internet Explorer.

I regret that even smaller fonts are not available for application other than Pocket IE. The Treo 700w (with push-email update) has a very good set of small fonts. It’s great for people who want to see more information at once. Other user can select larger fonts.

Phone
The primary function of the Moto Q is to be a phone, and as such, it behaves very well. All the phone functions work as expected and although some people did complain of the contact user interface, I did not see anything wrong with it.

Voice quality and volume are good, and pairing with a Bluetooth (BT) headset was easy and worked right away. The sound quality is also good when using Bluetooth. The downside of Bluetooth is that it depletes the battery quickly. During our test, our unit barely survived 8 hours of low-intensity usage (a day at the office) with BT on.

Camera
Not surprisingly, there’s a 1.3 Megapixel camera (with LED flash) in the back of the device. The still pictures are equivalent to a cheap entry-level camera, in good lighting conditions. The camera also captures 176×144 pixels video, which is smaller than anticipated. 320×200 would have the norm. The low-resolution makes the videos unimpressive.

PDA / Data
With the integrated QWERTY keyboard and an EV-DO connection the Motorola Q could be a dream email and web-browsing machine. Unfortunately, the potential for both tasks isn’t fully tapped into. For some reason, the EV-DO connection of our Motorola Q was only 250kbit/sec (tested on DSL Report’s mobile speed test in several geographical locations in the SF Bay Area). Our Treo 700w does 450kbit/sec under the same conditions and a PCMCIA card went up to 1.5Mbit/sec. We can definitely rule out the network and say that we expected much better from wireless data performance. This point will be a killer for many data-intensive users. I think that the 450kbits/sec of the Treo 700w is already not enough and I would be bugged to pay $39.99/mo to Verizon to get 250kbit/s.

For the test, we’ve configured the Q to sync with an exchange server, which is my favorite way to access email. Unlike the Treo 700w, there was no push-email available at publishing time, so the user can decide to sync every x minutes or do a manual sync. Push email is very convenient, but I think that most people can survive without it if they really like the phone. Pocket Outlook worked very well during the whole review.

I’m sure that a lot of people complain about the absence of WiFi. Well, yes, it’s true that having Wi-Fi would have been better (especially if it’s free), but honestly, if you have a decent EV-DO connection you might not care at all. The EV-DO coverage is much better then Wi-Fi’s and the EV-DO radio consumes less power too. Bluetooth is already killing battery life, I can only imagine how bad Wi-Fi would be.

Some users also raised the point that it’s not possible to edit Word or Excel documents. I think that doing real work on these types of files from a smart phone is a bit hardcore and not representative of most users. Personally, I could not care less about that.

Battery
The battery life is probably the weakest point of the Motorola Q. After all, Motorola had to cut back on a few things and the battery life was on the list. In practice, average users will have to charge it everyday. Heavy users will have to charge it each time they can: in the car, at the office and at home. Of course, there’s always the hope that someone will come out with a better battery and that would make life much easier.

Keyboard
The keyboard is slightly larger than the Treo 700w one, so if you have big fingers, you should check it out. Even after a week, I still prefer the touch of the Palm device. The Q keys are too thin and too pointy in my opinion. They remind me of my Treo 600. Both Palm and Blackberry have been improving on their keyboards and it seems odd that Motorola is a bit behind in this area. The keys are stiff, especially the function keys on each side of the directional pad. These two flat buttons would have benefited from being real buttons. The keyboard is backlit to a sufficient level to be used comfortably in the dark.

Windows Mobile
The operating system (OS) behaved very well during our week-long test. There were no crashes (phone hung) and in only a few instances responsiveness was low but overall, I’d say that this is more stable than my Palm-powered Treo 650.

The Motorola Q does not use the same Windows Mobile than the Treo 700w and the reason is: there’s no touch screen. However, the Windows Mobile 5.0 embedded in the Q is not as nice. Functions and settings are sometimes much harder to find and the “home” screen is fairly slow.

Although some settings are arcane, you should hopefully have to configure it only once and be done. The Bluetooth interface is a notable exception as it is very clear and simple to use. We have been able to sync with an external BT headset in a blink of an eye.

Another thing like I like with Windows Mobile (vs. Palm) is that I don’t need to install (and purchase) any additional applications. Web browsing, contacts, email are good enough right out of the box.

Performance
On the paper, the Moto Q looks very similar to the Treo 700w. In reality it is noticeably slower. That is probably the price to pay for the weight reduction. This translates into slower user interface response that can be annoying, but mostly OK. I believe that the reduced performance is worth the weight reduction.

Mini SD
In the Q, Motorola is using a mini-SD port instead of a SD/IO. It means that no SD/IO add-on like a GPS can be used with the Q. It doesn’t make much of a difference to me, except that I won’t be able to use my 1GB SD Card that’s in my Treo right now. Thankfully, mini-SD flash cards prices have declined steadily over the past year.

Conclusion
The Motorola Q is a slim, sexy and smart phone with a weak battery and a lack of true push e-mail. Luckily, both can be improved without having to wait for the next-generation Q. However, if you order today, that’s what you get.

Buyers can get it for as low as $150, which is very cheap compared to the Treo 700w alternative. Right out of the box, it’s great for heavy SMS users, thanks to the QWERTY keyboard. Data users wont be as pleased with performance or battery life, unfortunately.

In our opinion, this phone is the perfect device for a Phone/PDA+SMS usage.

Click here to check out photos!

Full Feature List

Right well, they did say they had a certain affinity for Seattle then, didn’t they? It looks like the first round of T-Mobile’s long awaited UMA service is now kicking around Seattle way, with open trial customers able to purchase two WiFi-capable UMA phones, as well as the D-Link router which enables home access. Right out of the gate we already have our qualms with the service — namely in that they’re charging customers an extra $20 per month for UMA access (and only if they have an eligible $40+ per month rate plan), which is kind of a bummer since in most cases UMA is cost-saving to the carrier in that consumers needn’t consume as much cellular air time. But that $20 extra also covers T-Mobile HotSpot access too, so if you’re not just planning on using HotSpot @Home, um, at home, then you’ll be good to go when roaming elsewhere. (It’s worth noting, however, that UMA does not require anything special to get going, and will work on any WiFi connection.) The D-Link router is probably a skipper — it’ll set you back $50 (and goes free with mail-in rebate), but isn’t necessary to use the service, and will probably leave you in pain as D-Link WiFi gear so often does. Oh, and hows about a word on the two phones you’ll be using with @Home: you’ve got the Nokia 6136, which, of course, features a 1.3 megapixel camera, FM radio, and GSM / GPRS / EDGE; then you’ve got the far more feeble Samsung t709, with GPRS, 176 x 220 display, and a 1.3 megapixel camera. Both will go for $50 with a two year, $100 with a one year.

Deutsche Telekom’s US cellular subsidiary, T-Mobile USA, has launched a converged fixed-mobile service that enables its cellular customers to make low cost calls via WiFi through home base stations and their broadband Internet connection, or through T-Mobile’s 8500 public hotspots in the USA.

In progress calls are handed over seamlessly from cellular to WiFi when the user moves into range of a WiFi base station and vice versa. However the tariff applied to the call wil be the one for the network on which it was initiated.

The service is a direct play for the nation’s fixed network customers with T-Mobile CEO, Robert Dotson, saying: “More people than ever are looking to drop their home landline phone and pocket the savings. However, they don’t want to use all their wireless minutes talking from home. Our new service solves this dilemma once and for all.”

The service is being launched with a choice of two WiFi enabled cellphones: the Samsung t409 and the Nokia 6086, both at $US49.99 with a two-year contract on a qualifying rate plan.

T-Mobile also has partnered with D-Link and Linksys for Wi-Fi routers for use with the service. It says these have been “designed for simple setup and enhanced battery life for the handset, as well as ensuring voice calls are carried with “the utmost call quality”. Each is currently offered at no charge with the HotSpot @Home service via mail-in rebate.

For a limited time the service can be added to any qualifying T-Mobile voice plan for $US10 per month for a single line, and $US20 per month for up to five lines on a FamilyTime plan. This gives up to five family members unlimited calling via the home WiFi hotspot.

I pulled this from wired.com today.  I thought it was interesting:

Despite its robust appearance, more than 10 percent of the internet flickers out like a candle every day, according to researchers who unveiled on Wednesday an experimental tool that probes the network’s dark places.

Ethan Katz-Bassett, a computer science Ph.D. candidate from the University of Washington introduced Hubble — a network of deep cyberspace probes scattered around the internet –- at the meeting of the North American Network Operator’s Group in Bellevue, Washington. For two weeks Hubble queried a sample of 1,500 internet prefixes (a small subsection of the net) every 15 minutes. In the end it found that 10 percent of those prefixes couldn’t be reached from certain corners of the internet.

Sometimes certain blocks of the internet weren’t reachable at all, Katz-Bassett reported, while other times only traffic coming from particular portions of the net fell into what’s called a “routing black hole.” When that happens, packets sent from one computer to another — whether a request for a web page, or an e-mail message — are somehow diverted to the wrong location, where they’re lost forever.

Harsha V. Madhyastha, Katz-Bassett’s partner in the project, said their tentative results surprised them.

“We’ve found a lot more reachability problems than we expected to see, with some prefixes being unreachable from several vantage points across multiple days,” Madhyastha told Wired News.

The researchers hope to build a tool that will chart these black holes in real time, by monitoring the dialogue that takes place between routers about the best path for particular traffic, and by building a permanent system of remote sensors that can send pings from various spots around the internet.

“A single unresponsive ping is likely to mean there are widespread problems, Katz-Bassett said. The larger system, which Katz-Bassett plans to build over the summer, would treat an unanswered ping as a canary in a coal mine, instantly triggering multiple probes from around the net.

Routing problems can be caused by a number of factors, ranging from problems with a particular router, often a new one, to ironic problems with a technique called “multi-homing” -– which supposed to make it easier for packets to reach their destination by allowing an internet site to simultaneously have a number of different addresses and network connections.

About 75 percent of the problems are fixed within an hour, and some last multiple days, according to their research.

Madhyastha and Katz-Bassett plan to make the Hubble data searchable by other researchers.

It finally has a release date: June 29. 26 days from now as I type this.

There’s nothing much more to add to the story for now. It’s still $499 (4GB) or $599 (8GB) with a two-year contract from AT&T (née Cingular). It still doesn’t have a 3G radio in it. And it still looks way cool. No news on what the monthly plan charges will be, as you’ll definitely need some kind of data plan to run all of its cool features. If you want some more up-close-and-personal action, Gizmodo has Apple’s three TV commercials for the iPhone (the last one is pretty amusing) that will begin playing any moment now.

How’s June 29 as an omen for a product launch?

Not much good to report, really. It’s the birthday of Gary Busey, Richard Lewis, and Fred “Gopher” Grandy. Much worse: Fatty Arbuckle, Katharine Hepburn, and Jayne Mansfield all died on that date, and the infamous Sampoong Department Store collapsed in Seoul on that date in 1995, killing 501 and injuring 937 more.
Bad omen or just unfortunate coincidences? I’m not sure, but I don’t think poor Fatty’s death will be enough to dampen enthusiasm for the “it” gizmo of 2007.

So… are you getting one?