What Comes Down Must Go Up


Over the years, I’ve only really touched on upload speeds with the National Broadband Network a few times, not because I have nothing to say on the subject, but because it’s so damn hard to explain the value to general users.

It’s easy enough to understand slow download speeds; a movie buffers (oh the RealPlayer days), your download takes hours, websites are slow to load or fail to load completely, there are many indicators to slow speed. Unfortunately with uploads many are at a loss to describe HOW they use uploads.

When it is explained that every photo uploaded to , video uploaded Youtube, or tweet sent, uses your upload bandwidth, there’s a sudden realisation that this may be something important.

And it is. To some.

For a general user, upload bandwidth is not as important as download bandwidth, just as download bandwidth is not as important as cheap internet. For a internet user like myself, high speeds in both directions are vital. If anything, I would be the kind of user that would go straight to 100/40Mbps NBN if I was ever to get it, which is looking unlikely at this stage: I have HFC out the front.

Then there’s business, where symmetrical connections are the bee’s knees, failing that, just a really fast connection with really fast uploads; say 1000/400Mbps for bigger businesses, 100/40Mbps for smaller.

Sometimes You Need To Be Giving

Upload speed (also called upstream bandwidth) is one of those concepts that doesn’t come into play until you start producing media for consumption. When consuming media you only use a small amount of your upload bandwidth.

A simplified example of a web page is you send a request to a “port” on a web server (port 80 or 443 usually), the web server acknowledges, sends data via the web server’s upstream connection. It’s not a complex system, but relies heavily on downstream bandwidth on the “remote end” (your end) and upstream bandwidth on the “local end” (the web server).

There’s heaps of sites out there to see some of this in action, a simple example is to look at the request/response headers on sites using something like web-sniffer.net. If you feel like hurting your brain, there’s plenty of sites out there with detailed information on how HTTP & TCP/IP deal with networking. You don’t need to do this, just take it from me that it’s a simple request-response system.

The problem for the remote end comes when they needs to actually upload something to for backup purposes, for example. I’ll use HFC as an example here, as it has the biggest disparity between downstream & upstream bandwidth (when there’s no load on the system that is).

Say you have your shiny new “100Mbps” HFC; it’s not really 100Mbps all the time, but that’s better than nothing eh? Not really. Anyway, you watch a few shows on Netflix, thumbing your nose at The Australian thinking you’re a pirate (YARR!).

You decide to upload some RAW photos from your DSLR camera, you need to keep them raw as this is a backup. You wouldn’t want to lose 30GB of photos due to an HDD failure (I have), so you decide to upload them to SkyDrive, Dropbox, Google Drive, or one of the other “cloud storage” providers. Surely a 100Mbps connection should be able to handle 8GB of photos you took over the weekend?

Nope. You’re limited to 3Mbps MAXIMUM  (UPDATE: I have been notified by a Telstra user that they do get 3Mbps upstream most of the time. One of the difficulties with Telstra’s HFC is that there is little in the way of solid documentation on their configuration) upload bandwidth on even Telstra’s fastest HFC. The actual speeds from user accounts seem to swing wildly between 256Kbps (and lower) and 1Mbps, topping out for some at 3Mbps. Even if you get the maximum speeds claimed by customers for uploads, that’s almost 6 hours, if you’re getting the lower speeds, that’s just over 3 days to upload 8GB.

Don’t forget, ADSL2+ tops out at just shy of 1Mbps, although it’s actually more consistent with uploads.

General users may not see a need for upload bandwidth, but there is one there. With the fidelity of media going up and up, we can’t discount the need for more upload bandwidth. Even my (cheap) little GoPro camera thrashes my connection if I want to upload video, and it’s not even shooting in 2.5K video, let alone 4K video (look it up, these are the next generation of video devices).

So users need it but don’t know about it, what about people who both need upstream bandwidth AND know about it?

web knotted cables

Business Case

One of the most common comments among my twitter compadres who run businesses is upload bandwidth. Everyone from graphic designers to server/network administrators all complain that there is just not enough upload bandwidth available to them with current connections.

This is not because they want to upload photos of their boozy weekends to , or want to share the video of a group of boozy teenagers taking “selfies” while embarrassing themselves. They need this bandwidth, and they know it.

What’s currently available, and their ONLY option under the new Coalition Broadband Network, is to find a provider that will offer high upload speed services.

Recently one friend, let’s call him George (yeh, it’s ), went through the painful exercise of having to find a provider that offers high upstream bandwidth. He was pushed to do this after Ziggy Switkowski announced he was destroying the NBN… wait, sorry, not deploying any more FTTP after end-of-year 2014.

So what were George’s options?

There’s HFC, which I’ve already stated is far from acceptable for uploads, as is ADSL2+, which is the second option.

The only two viable options George had was EFM (Ethernet First Mile) or a 10/10Mbps fibre connection (yes, that’s 10Mbps down, 10Mbps up).

Fibre really wasn’t an option; not only were the speeds half that of EFM services, but the costs prohibitive. $10 000 to haul the fibre and $800-$1100/month for a minimum of 3 years. Yes, that’s a $50 000 connection over 3 years just to increase the upload bandwidth by ~9Mbps.

EFM isn’t a new technology, it’s been around for since the early 90′s and is a proven concept that can be applied to DSL & fibre networks. Essentially it’s what allows Frankenstein style connections with split upload/download pairs, which is exactly what George ended up getting. The service is 20/20Mbps and costs him $330/month.

Even the most expensive 100/40Mbps NBN plans are less than 1/2 the cost of EFM, the cheapest 1/4, and that’s double the upstream bandwidth.

As we can see, there are just no options available to small/medium business users who desperately need upload bandwidth.

You may sit there and say “oh, they’re a business, they should suck it up”. Maybe, then again, I’m sure there are many people employed by small businesses whose jobs hinge on reducing costs inside the business. Think about it like this, $1100/month is a part-time employee. $330/month is an offsite backup service. $79.99/month is a coffee every day (at least). This is all without talking about the cost savings of moving servers back in-house (we’re talking about ~$15 000 a year there for a small business).

George sums the clusterfuck up nicely in an email he sent me:

So with Telstra/TPG connections, I’ll be up for $450/month, and via routing rules try to segment download/upload services to get the best out of faster Telstra download speeds and faster TPG upload speeds, and I STILL won’t be able to move my servers back, nor will I be getting the kind of speeds FTTH promises. This doesn’t even take into account that I have the technical expertise to set all this up, a cost that would have to be considered for a business with no in-house network specialist.


Back In Black

Sure, I’m a strong believer that businesses need to stand on their own two feet, but there is a limit to leg sweeps by telcos a business can handle. High set-up costs, poor support, ridiculously low speeds, and, most of all, exorbitant ongoing costs, mean that small & medium businesses have little chance of increasing their current upload speeds.

Even as an admin I found it painful to administer servers that were on ADSL2+ connections. I knew if I had to manage them I’d best get a coffee & find something to read. Simple account creations would take 30 minutes over Remote Desktop. You want a database backup? Call me tomorrow & I’ll let you know it’s finished.

The key is, when we decide which deployment type we go with (FTTN or FTTP), we can’t just look at download speeds. The best example I can give of ok download speeds and poor uploads is the old 1-way satellite connections of the mid-2000′s. 256Kbps down… dial-up speeds up (33.6Kbps).


In this day and age, the worst is, of course, HFC with the laughable “up to 2Mbps” that Telstra claims (or 3Mbps as users sometimes do get). Coming in a close second is VDSL2, the preferred technology for FTTN, with upload speeds being forsaken to boost download speeds.

Not only are VDSL2 upload speeds lower than even the EFM scenario I mentioned above, they do affect download speeds (this is just the nature of xDSL when deployed into an asymmetrical configuration). If you’re sending a lot of data, you can forget doing much on the internet.

Where does that leave Australians users & businesses? Well, up a certain creek without a paddle is the term I’d use.

There’s little for business to see as positive in the Coalition Broadband Network. Upload speeds will be lower than current systems, stability problems that existed with ADSL will still be there, and the options of having a business grade connection are still few and far between.

If you want to see a good visualisation of upload speed differences, check out the trusty How Fast Is The NBN by

Share: Share on RedditShare on Google+Digg thisShare on StumbleUponEmail this to someone
  • Friend_of_NBN

    The Estimates Committee on Broadband, particularly the meeting in Hobart supplied plenty of examples of the need for fast uploads. (http://polanimal.com.au/viewtopic.php?f=5&t=2551&start=50#p81918)

    Seriously—why the hell are the Libs destroying the NBN for? Is it the simian just smashing things Labor built? A deal with Murdoch to protect Foxtel from competition? Bit of both?