Not sure what you mean by "trunk bandwidth", it could be either backhaul or transit.

Backhaul (from customer to ISP POP) needs to match closely or exceed end customer bandwidth. Most last mile wholesale providers will offer a 1:1 or low contention SLA, although some "manage" the bandwidth (they oversubscribe until someone complains). However, if an ISP can't get this, or they build it themselves, they need to make sure your backhaul interconnects are overbuilt enough to deal with expected growth in line with however long it would take to build additional capacity. This is the primary cost of end user connectivity. The local loop price is usually bundled in with backhaul for costing.

Transit is cheaper and much easier to manage, these are your interconnects to other networks. It needs to be overbuilt as well to exceed peak usage by enough margin to allow continual upgrades, but aggregate usage across an ISP is generally much lower than the sum of its customers. Transit physical interconnects are usually delivered directly into an ISP's POP (a location commonly shared with other network users and ISPs, like a regional datacentre) and benefit from economy of scale.

Discussing the cost of pure transit isn't too useful. Heavy users smash both backhaul and transit if they go nuts, and this usage can be much harder to predict and manage than 1000+ users calmly watching youtube or emailing kids for 5GB/mo. It is much easier at large scale, but for a small, non-profit rural co-op I imagine the handful of big users they have could get very expensive.

In the pricing model, there needs to be an element of discouragement (suddenly heavy users with deep pockets can still degrade network quality for everyone else) as well as recouping the cost of required upgrades and improvements to support the traffic. It is ridiculous for a non-profit to massively overcharge on the scale you're suggesting for no reason.

I'm not sure what Net Neutrality from the summary has to do with any of this either - this is usage-based billing over a flat pipe, not charging/throttling based on traffic type or destination. I may be a bit biased as I'm from somewhere where quota plans are the norm.

Getting a bit OT here, but I have not worked at a single company where the CEO/Managing Director/whatever did not work at least 2x the number of hours of practically everyone else.

For my current boss, stock market dabbling is leisure. Wining and dining whiners and strategic customers can be fun but it means he doesn't get spend time with the wife or golfing or just chilling in front of the TV. He's in at 5am checking projections and talking with vendors/big customers, regularily leaves at 4pm to go to business and networking seminars until late at night, or is just in the office until 6-7pm.

He's in his sixties, this is an established business that's been around for decades. Would you have the energy to build something like that from the ground up? I don't. He did. If he wants to relax a bit and drop the average back to a low 70-odd hours a week, good for him.

What?

D-Bus is still portable across multiple free Unices and even Windows. The standalone daemon isn't going away anytime soon, and I can't see the multitude of projects depending on it giving up cross-platform compatibility.

In-kernel IPC reduces context switching and other related overheads. I'm not sure exactly how much of a performance gain this gives D-Bus clients and the system as a whole but if someone wants to spend their Christmas playing around and seeing if something is better then great! And that's part of the benefit of OSS.

I thought this was already the case.

At least here (AU), it's been practically impossible to get a MasterCard or Visa-backed card without a smartchip for half a decade, and in 2014 signatures will no longer be accepted to validate identity on credit purchases. There's been ads running for about a year requesting that people create a PIN for each of their cards. (AFR rundown)

Bank-issued cards (not store cards) always come with NFC as well now (doesn't seem to be any way to request otherwise). The last non-NFC card I had just expired and was replaced with a Visa PayWave. NFC & RFID is also very popular for specialist stuff : cabcharge cards, fuel cards, public transport.

A few times in the past I've done up quick spreadsheets to compare buying food vs. eating out. I'm in the same boat as a lot of others in the this thread - poor diet, long hours, fast food. I'm not going to pull stats out because they keep changing, but as general trends go:

Eating well on restaurant food is silly expensive. Noone is arguing about this.

Buying and cooking good food vs eating fast food is a close tie. I'm not a huge fan of veggies and meat/pasta is expensive to make up the bulk of your diet, so YMMV.

Buying and cooking crap food *can* be cheaper, if you're not buying microwave pizzas and burgers.

All of this is before taking into account the cost of my time. Once that's taken into account, it makes dropping through an expensive "healthy gourmet" carvery with free wifi or decent 3G coverage a hell of a lot cheaper than spending 30 mins a day preparing lunch and dinner. Eating out for lunch and dinner means no mess, I can still bring my laptop in and bill my time while I get off my feet, away from customers and enjoy food that's far better than I can cram into my schedule if I'm making it. I get to sleep an extra 30 minutes! That's worth a lot more than the money.

Everyone's scenario is different. I try to eat good food if possible and avoid the golden arches like the plague. Sometimes, though, you end up working 16 hours in a day without a break, it's midnight and they're all that's open. It beats scrambled eggs on toast. If you have a nice 9-5 job that you enjoy and can spend your weekends pursuing your shaved meat hobby, well, that's great too.

And yes, I take multivitamins to back up my highly variable & mediocre diet because it makes me feel healthier. I'm completely unworried that it might be just in my head - as long as I'm not taking relabelled rat poison and they're cheap as chips, I get a benefit out of it so they're worthwhile.

NBN doesn't use compatible GigE technology, so you wouldn't be able to use an fibre SFP or switch anyway. NBN's even heavily customised the firmware on the NTUs so nothing is quite standard.

The biggest issue with HFC is the shared medium. NBNCo fibre uses a 2.5/1.2Gbit OLT with a 32 or 64-split GPON local loop, a design that shares many of the same issues and has a maximum design speed compariable with FTTN w/VDSL local loops (~100Mbit). The biggest benefit of fibre is being able to deliver 100Mbit over 20KMs instead of 300m with DSL technology.

Intel are high-midrange (in terms of quality and performance) in both enterprise and desktop flash storage. Samsung has a very strong presence across the board due to reputation, very good performance (substantially better than Intel in production gear) and many of the vendor-branded enterprise drives (HP, NetApp, EMC, etc) have Samsung internals.

When I bought my current desktop's SSD (Samsung 840 Pro), the only other vendor with a drive even remotely close in performance was OCZ. Googling for "OCZ Vertex4 problems" quickly put that possibility to rest. At the time, Intel wasn't looking like they were bothering to keep up with performance on the desktop, but they're always been reasonable quality-wise.

Or the strongest traded currency in the world, the Euro. It's also fairly popular for trade between the biggest energy exporter in the world (Russia) and the Eurozone. OPEC has several times publicly announced a desire to convert their cash reserves into euro.

True, it only has a small lead by worldwide trade value and runs a close second behind USD on trade volume, but even with Europe looking a bit shaky, the USA has its own problems.

There's a strong and escalating push outside of the US for nations to agree on alternatives to USD-based trade agreements - China has joined Europe and Russia in cheerleading these efforts.

10 years ago if I ordered wholesale goods from China, I would pay in USD-denominated amounts. Nowadays, the same factories and vendors accept even direct AUD payments and will invoice in my currency of choice.

That's across the lowest-latency, highest-performing set of international backbones out of Australia.

Now try a traceroute to waia.asn.au, hosted in Perth WAIX. I get 80-100ms from my core router in Brisbane, which has multiple carrier-grade fibre interconnects to upstream providers. A residential connection will add another 30-50ms. Perth is only ~4500kms away as the cable runs. You would be slightly better in Sydney (by ~10-15ms).

Some ISPs have poorly organised routing and peering arrangements that (for e.g.) terminate all VCs at a state-wide ag in the capital, so users next door to each other in Mackay have to route via Brisbane to ping each other. A Brisbane user trying to access something on a UQ website might end up going to Sydney and back instead of across an IX in Brisbane, which adds another ~30ms. Now imagine roaming users on 3G or LTE trying to use Citrix with latency bouncing up and down due to the medium itself.

A lot of what I do is either application delivery (Citrix, RDS, SSH, etc) or voice-grade comms. Our entire WAN business is built off the back of our experience in that area. 100ms is annoying. 200ms is unusable, from the perspective of Jane Sixpack sitting in front of her Citrix ERP app trying to key in phone orders or look up customer details. Jitter is incredibly frustrating and the source of many grey hairs.

International just gets silly. Average latency from AU to USA is 150-200ms, to the UK, 300-500ms, to Russia or Scandinavia, 500-600ms.

I'm not sure where you got this information from, because a workmate has just spent a month migrating a large organisation away from 365 and back to local Office/Exchange/etc.

Office 2013 doesn't require the cloud and works fine with local files. Exchange 365 supports direct mailbox migration to another Exchange 2013 server across the wire (no dumping to PST), Outlook just picks up the new settings without an issue. He ended up synching file stores to a common, DFS-R replicated location. There was a surprisingly minimum amount of fuss. Most of the problem was slow transfer speeds (mailboxes of unusual size) and users griping about how long it was taking.

I'm trying to find which Lumia you're talking about and I can't. The closest I can get is the 810, which was discontinued after about 6 months (still pretty short), presumably to make way for the 1020.

Say what you want about the phones, everyone I know who's purchased a Windows Phone has liked it and most of those are Lumias. Admittedly, there's not a lot of people I know who have WinPhones.

Considering that every phone I've purchased for myself has been a Nokia (still on an old N8), I'm seriously considering a Lumia myself. My work phone is an iPhone so I'm not keen on having 2 of them competing for which battery runs down first.

If anyone would've suggested to me that WP would've been an option a couple of years back they'd have been crowned the world's greatest comedian.

Unfortunately that's exactly what the business market is doing. It starts with spreadsheets and Citrix, the occasional email. ERP vendors are now rolling out apps to use them as simple endpoints either in the field or for simple tasks like stock control.

I've got a few customers using either iPads or Android tablets with barcode reader attachments to remove the need for paper records in their warehousing and logistics depts.

There's a lot of areas tablets work well. The problem is businesses trying to flog them off as a primary device for all users. A Surface Pro with a better battery might be able to replace a laptop because it *is* a super skinny laptop missing a keyboard. iPads, Surface RT, even the ASUS Transformer, not even close.

It's 40 million in revenue, not net profit. They're charging $400 per advertiser on average, if the numbers are accurate.

Alternatively, an $80/mo Linode (or similar) plan would cache 2 days of data (~200GB storage), offer some capacity to 'cook' it a bit before re-downloading (say, do some compression) and have enough transfer (8TB/mo) all in one shiny package. For pure storage, I think Dropbox and similar AWS-hosted services weigh in around the $60/mo mark at what would be needed.

Personally, I would spend money on an additional, dedicated Internet connection or (better) WAN tail to the customer and drop some staging hardware on their network border to ensure outages don't result in lost data.

The words 'budget' and 'cloud' together usually result in a selection of four-letter words, most notably, 'pain'.