None of these protect against a volume-oriented DDoS. Many are DoS only (single / few sources) and do not apply when every IP on the Internet appears to be sending thousands of requests, or more likely, responses. Further, you've completely ignored spoofing of addresses combined with amplification attacks (send out a 64 byte DNS request pretending to be the DDoS target, get 4kB sent to the target). Finally, regardless of the 50-100Gbps pipes MS, Sony and Amazon no doubt have, they're useless when there's 1Tbps of amplified crap directed down the pipes. With the example above, you'd only need about 4Gbps of bandwidth total (40 cheap VPS on "100Mbps" connections) to generate 256Gbps of DDoS.

When 256Gbps of rubbish arrives at your servers or firewalls ... registry settings and kernel tweaks do jack (note that CloudFlare was hit 11 months ago with more than 400Gbps of DDoS, so this is not implausible!)

And since it seems it was apk I'm replying to ... I'm actually half surprised you didn't try to claim that a HOSTS file would magically help.

But would you also take JUST Fox News for $20 a month, with no other channels, or Fox + 20 channels of crap for $10 a month? Because that's the type of "offer" the content networks generally make.

Buddy, you can get a certificate for less than FIVE US dollars per year. Is that too much for you?

Actually yes, frankly it is. Because according to Google's overpaid, brain-dead Chrome developers, I need one for the KVM, one for each of the management cards in the servers, one for each of the appliances I have (from DVRs to firewalls etc), one for each little device with a web server (assuming it even supports writing a certificate to storage, and config for HTTPS), one for each workstation or server with an app or config UI. Quick count for my house alone ... 47 certs excluding the devices that quite literally have NO way to store and use a cert. I simplified too by assuming the devices supporting certs can handle SHA256 (thanks Google for THAT little recent shitfight). And the certs don't support SANs nor do CAs allow local names, so I have to use the correct FQDN all the time now (no more http://dvr/ or typing the IP - now it's https://dvr.private.example.co...). And what have I gained? I've had to spend $230+ and several hours of work to avoid irrelevant anti-sec warnings, on devices no-one can get to except me. It's bulldust.

And that adequately reflects the rest of the world how? I have customers with multiple 5Mbps connections (literally the best they can get, there IS NO FIBER) at $400/month. They have dozens of users, 10-100MB files to send and receive, every day, and therefore a local caching proxy is the only way they can get any reasonable web access at all. But go on believing the rest of the world is like your little Utopia.

No - this problem is solved with SNI (Server Name Indication) which is part of all the current browsers, and has been for a while now. The client tells the server which certificate to return (which hostname it's going to ask for) in plaintext. There's probably a module you need for Apache to support this - IIS finally does it natively, so I'm sure it was already there in Apache/nginx.

And forcing SSL does nothing to prevent your employer setting up an SSL proxy with a wildcard certificate, decrypting everything you request, and tracking you anyway. I've set up MITM proxies for companies before, and it's literally 10 minutes of effort in most cases (because the end-users already trust the corporate CA). And if you think the Government can't MITM you as well you haven't been paying attention for the last 12 months.

OK, Mr AC, care to explain how you plan to cache SSL-encrypted objects? All your caching proxy sees is the "connect me securely to server X" request - after that, it's encrypted and your proxy cannot tell what's being loaded. Worse, since SSL inflates the data sizes of whatever you've requested, your images are up to 50% more data, and your (already compressed with gzip) HTML, CSS, JS etc is the same. So you've added 50% to your traffic for ... potentially nothing.

Seriously, what do you gain (actual, measurable improvements) from switching from http://www.comics.com/garfield... to https://www.comics.com/garfiel...? Nothing but overhead.

And that's leaving aside the fact that SSL no longer guarantees the source server (too many options for MITM server certificate hacks) or security (POODLE etc).

No, make no mistake, this is Google throwing its weight around, screw anybody who doesn't want or need a certificate for their site, or has made a conscious decision NOT to use SSL (not to mention all the corporates with proxies that inspect for malware - now you're mandating SSL MITM by the organisation, or you have a channel for malware into any system).

Ah yes, one bad patch and we should all NEVER PATCH AGAIN BECAUSE THE SKY IS FALLING! Perhaps he will take personal responsibility the next time a patched vulnerability launches a new botnet? Nah, just write inflammatory rubbish, it's easier.

Actually, I don't know why they don't "acquiesce" somewhat to the demands - and offer to sell to the dealers at the same price as they sell in other states.

When the dealers refuse on the basis they won't be competitive with out-of-state sales, they should surely be able to use that to force the hand of the legislature (by advertising in Texas, with the tag line "Not available in Texas because none of your dealers will sell our cars" or something). Truthful. Pins the "blame" where it belongs (the dealers).

If, OTOH the dealers accept, the customers will demand to know why Texas is 25% more expensive (and Tesla can truthfully say "We sell at the same price to all comers, dealer or private, so any difference is the dealer's margin because your state gov't won't let us sell direct to you".

I'm very interested, with Tesla apparently coming to Oz next year, to see what happens here.

Or you could read the article (psht this is SLASHDOT, what was I thinking?) and the papers it references which indicate the most likely outcome of an explosion of the craft within 1m of takeoff would still result in 0 deaths. Science, not baseless assertions.

While it's true that it takes months or years for the number to be re-issued, it takes only an hour for it not to be your number any more after you change providers (or, in the US perhaps even area codes?) In Aus we have number portability between the carriers, which is nice when you pay for it - but sometimes you have to change numbers for reasons outside your own control. I trust (from some of the above comments) that this new tool handles what would seem to be a fairly regular occurrence, though the summary suggests otherwise?

Sure - I could. But that's extra devices and usually extra power points at those locations (esp if you want any POE - I doubt there will ever be a switch that can be powered by, AND deliver POE at the same time). So it's extra devices to buy and support and manage which is why I decided against it. Having the extra ports doesn't stop me doing it in the future either.

The flip side of course is that a failure in one of the big switches takes a LOT of things offline and it's more expensive to replace. Not the VM cluster or servers - but about half the other devices (e.g. one of the WAPs, half the desktop points etc).

OK Telstra has to record the source and destination numbers of all the calls - right? Here's a sample record (not that drawing a table is easy so work with CSV here):

FromID, ToID, TimeStart, TimeEnd
0299999999, 0288888888, 20090617135834, 20090617140711

How would you like to determine whether the number 0299999999, which is not owned or operated by Telstra today, and which was not owned or operated by Telstra in 2009 either, was or was not an unlisted number at the time of the call? Because its state right now is completely irrelevant - the state at the time of the call is the important and relevant piece of data, and it doesn't exist. And the reason it doesn't exist is that this is a record designed for billing and cross-checking, not for customer view (if you're arguing against unlisted numbers in toto, you've never been stalked).

I did this when I finally bought a place 15m ago. I went what I considered was pretty "nuts" on the cabling. Cat6A everywhere - 2 in every room except bathrooms, kitchen, laundry and foyer, 6 per room for the entertainment areas. 2 APs at opposite ends of the house, and everything terminates in a 6U cabinet in the garage (26 points total). The sparkie who did the cabling said he's just finished another place with over 50 points, similar approach to mine. So what would I do differently? Most rooms are fine. I find I could use more in one of the entertainment areas, but some of those devices are both wired and wireless (and if push came to shove, I would simply move a device to WiFi). I wish I had thought to put a couple of points near where the solar inverter will be, so I could run a Galileo or similar for monitoring - it'll have to be WiFi. But this gives me at least 1Gb with POE almost everywhere, and I can go to 10Gb if it's ever a requirement.

Look I know it's a tiny thing, and I'm in the "don't like U2 so might have been annoyed" camp. But at least some of the reasoning behind the annoyance is that this has hit a stack of data caps / data plans on mobile devices. "It's only 100MB" you say. But if that's 1/5th your monthly data and you only had 30MB left on the last 2 days of your month - now you have a bill thanks to Apple. And where does it stop? "Here's your free 100MB download" is a possible annoyance or a great thing once. It's a royal PITA for lots of people if it starts being every month or week. Or what if it was a 1GB movie instead? Is that OK because the free 100MB album push was OK, and $producer paid Apple eleventy squillion bucks, and it's free so don't complain? Sorry, there's nuances here you're deliberately ignoring, and it makes your argument look like a baseless whinge.