Then you're not paying attention; some already have.
Agreed. In every discussion I've had with customers about IaaS and cloud, the security aspect has been the #1 topic of conversation brought up by the customer. Closely followed by performance.
Businesses of all sizes and industries are very interested in all this mess in the cloud space.
One of my first IT jobs was working under an ex-Army sergeant IT manager. He used to work on Prime mainframes for the ADF before going corporate, and had a lot of experience with project management on the high end of defense & government IT.
We all had an immense amount of respect for him, whether it was me (early twenties) to the older 40/50-odd techs. If you screwed up, you got chewed out. He did a pretty decent disappointed-dad routine which was somehow worse than being yelled at.
We respected him because we knew he gave a crap and kept the "Directors of Micro-Management" well and truly off our backs for the most part, made sure Technical Services was involved in major decisions as much as possible. We would get warned if something nasty was trickling down from the top. It fostered a healthy team spirit too, everyone looked out for everyone else, helped each other with problems and so on. If a customer was causing problems or making spurious complaints, our manager would be out there looking them in the eye and asking if they had a business case for being difficult. On the other hand, if he thought an issue was your fault, you would know in very clear detail the hows and whys.
I only left that job when the company was acquired by another much larger one and the TSM was made redundant - though that's a much more involved story.
Ever since then, I've sought out companies with a similar "feel" to work for, and tried to foster a similar tone when taking up management roles. I've been successful finding places to work, and I'd like to think I've done as good a job as I can for the rest.
What's the point of this TL;DR? The biggest thing I've noticed is attitude makes up most of your workplace perception. I've worked in crappy places since (though not for long). Keep your head and chipper helpful attitude, useful people notice and come to you for direction instead of your boss and the vibe spreads. Walk around with a martyr complex and no-one will want to help. A healthy team means even if it's busy, it's not hard, and you have a lot of potential extra hands that may be more skilled than you. This goes a long way towards getting a recalcitrant boss on your side too, if they have any kind of management skill at all.
Can't really help with bad management, if they don't recognise your value, leave. You'll find out if you've overestimated on that part pretty quickly.
Since we're talking business users, VBscript macros. I know there's some interoperability but it's far from perfect.
Lets think of some other show-stoppers:
Integration with pretty much any DMS.
SharePoint (very popular as a cheap DMS, especially the free one). You can use it was a web-based folder store but it's a pain in the bum to perform workflow operations, rollbacks or checking change sets.
Document workflow add-ons, like PDFComplete and Nitro, that accountants and legal firms large and small commonly use to cryptographically watermark PDFs and electronically track paper trails (often integrating with some sort of DMS backend).
Access application support. This is hard even between Access service packs. There's a large number of hand-me-down Access applications within SMEs that noone can access the backend of anymore but noone wants to rewrite after Bob the Tinkerer left the company.
Full Exchange integration. This is the biggest upsell I've come across in convincing people to move to SBS/Windows Server & MSOffice environments. Simply being able to share calendars and mailboxes without dodgy third-party PST hacking add-ons or having to deal with IMAP/iCal makes small business owners' eyes go round with wonder.
All of these things I've had to deal with for new customers within the last 3 years. There's many more examples than this. Usually these customers have grown out of being supportable by tiny 1 or 2 person IT shops or they've recently been stung by trying to go for the cheapest, poorly designed and implemented option tabled by some buzzword-driven cowboys. The amazing thing is, 1-2 years down the track, staff are happier and they've spent much less on support.
Years ago, working in education IT in my teens and early twenties, I spent a lot of time pushing LTSP, the K12 distros, OOo and friends for smaller schools (older computers, perfect for thin clients) and kindergartens. There was a bit of resistance due to unfamiliarity in some situations but us IT guys were pretty switched on with a rounded enough skill set to build and maintain a solution that worked in that kind of environment.
Small companies don't have a large, dedicated, government-funded IT crew on tap, and they can't afford $160/hr every time someone loses access to a calendar. They're very fast-moving compared to educational IT with highly varied requirements.
In most cases, Office "Just Works". When it doesn't, the 18-year-old helpdesk guy can usually fix it. You don't need to waste the time of a senior resource, or bill the customer for his time. For most businesses, having less headaches is well worth the additional cost of licensing.
I would be stoked if I could design in FOSS components into the solutions we support. We're trialled various components many times in the past. But I can't be confident they will work for and with all the requirements, and most importantly, that I can get resolutions for support issues immediately. Even with a support contract, outside of a few huge OSS powerhouses (eg, RedHat), support responsiveness and available skill pool leaves a lot to be desired.
Really? Looks to me like it's gone down consistently over the years, looking at past pricing on products since around 2003. The biggest anomaly in that is the abolition of Small Business Server resulting in a 20% markup on SBS/SBS Premium functionality after June this year.
Per-core licensing of SQL 2012 is (a fair bit) cheaper over 25-odd users (depending on feature set), SQL Express is free and each new version contains even more of the add-ons you used to have to pay for, Windows 7 & 8 was and remains much cheaper than XP was this time from release, Office has come down a little in price at Pro edition and a lot at the lower end editions or singleton products, Visual Studio is a bit less ridiculous and even has a free version, Server 2012 Std licensing offers effectively a superset of Server 2008R2 Enterprise's functionality at a slight increase over 2008R2 Standard pricing (there is no equivalent to 2008R2 Std anymore).
I could go on a bit more. There's been a lot of pressure from high-quality open source or lower cost alternatives, so MS has been pushed to show a bit more value.
Big licensing upgrades are admittedly horrifying to customers, when they see a $30,000 line item and ask what it is, and we say, "Oh, just Office". It's at the point where hardware is so cheap it has trouble making 1/3rd of the cost of a project, and it gets worse the bigger the organisation.
Turning off the key loses brakes and steering, if it even turns the car off. I'm pretty sure it's a push button start.
I'm also guessing that if the computer is really malfunctioning enough to ignore input from the accelerator it might also be ignoring input from the gear shift and start button. If someone has an hours worth of time and the mental capacity to call in a police escort they've probably tried a few things before getting to that stage.
The best bit of the article is this:
A Renault technician had been on the phone with police throughout the chase trying to help but couldn't come up with a solution.
Snort and associated tools aren't too bad, and should run on most Linux/BSD-based custom firmwares if the hardware has enough juice. A Cisco ASA with an IDS module is less good, but servicable. You'd need to use ASDM for monitoring unless you want to buy a super expensive monitoring suite.
The main issue is the amount of processing power and RAM required, especially if you're pumping through a lot of traffic. I run pfSense in ESXi on a little HP Microserver as my router. Using default settings with Snort, pushing through ~20Mbit of WAN traffic makes the poor little box bulge at the seams. It can be tweaked and the rulesets rejigged or disabled, but for Snort it's a RAM vs CPU tradeoff, and disabling rules reduces the level of protection. There is no way a cheap consumer router could have the resources to do it properly.
You're looking at more the SonicWall (ugh)/WatchGuard/NetGear/etc SMB/UTM router space. None of them qualifies as anything other than "terribly painful to use" for IDS/IPS.
If someone is using an iPhone, at some point it was connected to iTunes to activate it (or it wouldn't be working). The vast majority of people I'm aware of - technical or not - who have an iPhone also use iTunes, even if it's only occasionally on the work PC or at a friend's house. I had elderly relatives of friends asking me if I'd tried iOS 6, and showing me all the cutesy new features before I'd gotten around to upgrading myself. My iPhone is a work phone so I tend not to jump on updates immediately.
iOS updates over the air since 5.0.1, with a big obvious blob on your settings button if you don't, and nags every time you plug it into an iTunes-capable computer.
It can still be handy to use a speed test attached directly to the provider's core network to determine where the problem is - my home ISP has a few large files on a customer website you can download, and their customer support will get you to try both an internal and external test site.
If the internal test is slow, then there's a problem on the backhaul between you and the ISP. If it works but external sites are slow, there's either congestion on the edge (usually pretty obvious on a traceroute, and your ISP should be very aware of such problems) or issues beyond the ISP's control. Still, getting traceroutes of such things allows them to pull their transit and upstream providers into line.
It's fairly common to use NetFlows or similar protocols to measure per-IP bandwidth at either the inner or outer edges of the network (depending on whether it's billed over all traffic or just external traffic to the ISP). NetFlows has a few issues where it under- or over-reports the traffic used by a network flow due to insufficient detail supplied by the protocol.
In most cases where an LNS is in use (such as with any PPPoX DSL connection), they can report down-to-the-byte accurate statistics, but often don't - it's easier for an ISP to centralise everything on NetFlows, which works fine for any end-user tail.
We have had a few problems with upstream providers giving us huge bills for specific customers, and us going back with a handful of graphs and spreadsheets to prove that their stats showing 50Mbit/s usage on a 20Mbit/s DSL circuit is highly improbable within the confines of modern physics, our monitoring of interface stats showing the same packet count with a far smaller payload.
I've had a couple of interviews like this. It really is a fantastic thing.
Once of those interviews did in fact end up getting me a job by word of mouth back in my younger days, it was completely worth it. Fast forward to today where it *is* a big part of my job to do this for people - as a presales solutions architect.
I did a double take reading TFS for this article, I can't imagine thinking like that. If you're asked a reasonable job-related question in an interview, you answer it. If you can solve one problem for them, it's likely you can solve more. If they're opening interviews to bring in outside opinions, all you've wasted is time. Someone who doesn't want to hire you or pay you for your labour obviously isn't worth working for, and if they get an hour or 2 of free successful consulting out of you, well, they're losing out if they don't hire someone who can pick up the state of their business and nut out a workable solution in that amount of time. I'd go find somewhere better, and a minute shift in attitude might help find that dream job.
As a Canadian, I like the connection we have to Australia through being in the Commonwealth, and never saw why so many of you guys got bent out of shape over what is really just a figurehead...
They've tried a few times and failed. There's no huge push for becoming a republic, the opinion polls are pretty even, and the last referendum was 55% against on all motions. Most people I know are pretty ambivalent about it, not really worried either way.
Me personally, I don't believe changing the status quo benefits anyone but the politicians. We have a historical connection to Britain. Our parliamentary and legal authority is vested in our own institutions. I'm somewhat wary of letting our pollies re-write and amend big chunks of our constitution to allow our head of state to use a different title on the departmental letterhead.
The view that the kernel has of memory allocations is somewhat different to what the application sees - most of the work is done in the libc, which directs the kernel to map regions and size the heaps that it malloc()'s from.
Usually, you'd have GC handled by other libraries, for instance the Boehm-Weiser GC. If you're using C++, Boost also has a few wrappers and implementations of garbage collection algorithms for a variety of use cases.
From memory I believe the kernel CMA is mostly related to in-kernel allocations, so drivers and kernel threads.
It sounds like a peering arrangement turning into transit. There was a comment earlier about extra detail from French language reports on this story; Google is trying to break into certain African markets that Orange have a large presence in. Orange, like many large carriers, is having a hissy fit over the traffic volume coming from Google services. This would be an olive branch from Google being used as a back-scratcher.
It is not uncommon for Google to co-lo equipment and caches in the datacentres of larger national carriers. Usually this is peered directly or hosted by a common peering IX. Google in this case may have agreed to pay directly for transit.
It's difficult to tell exactly what is going on without more info, so far we have a bragging CEO and no real solid detail.
I'm wondering if there'll be any announcements about the Gertboard and similar prototyping tools for the Pi? It's a shame it went off sale about a week before I received my Pi.
Striking: a combination of denying your services (and only yours) to an employer with (usually) highly visible protests and PR to get the reasons for the strike out to the general public, thus gaining their support and pressuring politicians. In most locations, it is illegal to actively interfere with the employers' ability to do business beyond withdrawing your own labour and expertise - blocking access to sites, harassing customers, etc will probably attract police attention. If a group of other people decide to do the same, they've done it under their own steam and should not be coerced to do so.
DDoS: actively suppressing the target's ability to communicate and do business with very little cost or effort, and a high potential for serious collateral damage to the operations of completely unrelated individuals and businesses along the way. In some rare instances individual participants may volunteer their resources to the DDoS, but much more often a lot of traffic is coming from compromised systems. DDoS and systems intrusion are both already criminalised actions.
I'm 100% for the ability to protest or strike - I've done my share in the past. I'm 100% against bored kids getting their mates together to smash small websites into gravel "for the lulz".
You can argue that it'll only be used as a responsible form of protest against those that deserve it, at which time my eyebrows will climb my forehead and burrow under my hairline. The amount of effort required to "protest" is much too small, the relative damage against small operations and individuals far out of proportion. If we would like to see spam outfits "protesting" against Spamhaus' DNSBL resolvers, pro-X advocates "protesting" against the blogs of their pro-Y opponents, kids targetting the local donut shop because they're bored and want to give them a huge bandwidth bill, sure, lets allow legal DDoS. Perhaps we could allow governments to set rules on what constitutes "fair protest" to solve this problem? I'm not really warming to that idea either.
They all use slightly different versions of libraries, each of which introduces slightly different bugs and issues into the environment.
As long as you have the right dependencies on the package, its all gravy. packages are just renamed compressed archives, and they can easily be repackaged in 20 lines of bash, something that runs on all linux machines.
You still need a deployment build environment and a test platform for every target - which would include every sub-distro you want to support, as they all have different package sets right down to libc, every "branch" (testing, stable, etc) and so on.
It's quite difficult to package games using the native package manager in any distro and expect the package to have any kind of supportability or longevity. If the vendor has to support it over time, they're only going to support a select few distros, or they'll be enjoying stress headaches until support is withdrawn.
The easiest options for binary software distribution are what we've seen many times before: statically link the whole lot (or at least remove external dependancies) and distribute a tarball (like id did with all its Linux releases), or if you want to get fancy, an installer script/GUI like VMware or WordPerfect (back in the day). This can be wrapped up in a native package but it's not going to be adding much value beyond showing up in the programs list.
SBS is discontinued past SBS2011 - you just have to buy the separate products and pay for them unbundled (about a 15% increase in price on a project, depending on your CAL model and quantity - going on a 20 seat basis here). Server Essentials 2012 is something else that's been around for a while with one name or another.
You're entirely correct about them pushing the cloud model with these changes - we've had a bit of pressure coming down the channel to push Azure and 365, which has been met on our end with a mixture of quiet sniggering and outright mirth.
A big part of my job (at an IT integrator) is going out with salespeople to chat with customers when potential projects are looming, determining requirements and a suitable solution. From that perspective, we've had a lot of questions asked about Windows 8, so we've run some demo units up to loan out for the curious. Our help desk also had a couple of their workstations converted to familiarise themselves with the environment.
Within a business environment, the feedback has been entirely negative. We have yet to do a Win8 rollout within a client. Internally, those deskbound staff who got a Win8 PC have since reverted to Windows 7 after trying really hard for a week or 2 - I'm not sure why, it more likely relates to us having a lot more clients running Win7 and helpdesk staff wanted to have an environment to match. There is still a spare Win8 demo PC for internal use by any tech team members. Even reverting to a "classic Aero" interface with a start menu was frustrating.
Our customers found the interface very confusing (far more so than WinXP to Vista). Even with classic mode, they did not like the distractions, the interfaces, and felt that nothing in Windows 8 added business value beyond what was available in Windows 7.
All of that is pretty obvious - I never expected something touted as effectively "Windows Touch Screen Edition" to have a huge business focus. However, I've been talking to our various IT contacts at customers - we would talk to a few hundred at least once a month, managed services and whatnot - curious about Win8 for their own use on laptops, tablets and home PCs. I gave them my opinion of it (which isn't high from my admittedly limited exposure). The feedback there has been almost entirely negative. Even customers who adore their new Windows Mobile devices hate the similar interface in Win8. I don't think we've sold a single PC without downgrading to Win7 first.
Very few of these people have used anything but Windows since '00. We HAVE had a big spike on MacBook purchases recently.
Admittedly, we're an integrator and we do not box drop or sell retail product. Harvey Norman might have a different story, but they don't have to deal with the support issues on the back end. We get that when a CFO calls us because his kids can't use the new home PC - and we're not going to tell him we can't help.
Consider that a lot of central phone exchanges and DCs are in the Brisbane CBD, and most kept working in early 2011 when a good chunk of the state's land area was underwater: http://forums.whirlpool.net.au/archive/1616757. The locations I've got gear (PIPE DC3, Pegasus and Fujitsu) were well out of the flood zone, being located in areas above any reasonable flood level - if they had gone underwater, there wouldn't be a Brisbane anymore.
The main drivers of DC location are cost and utility. Cost includes the normal stuff like power, real estate, connectivity and so on, but also factors in risk, redundancy, insurance and mitigation of each of these factors. Even though Sydney is a relatively low risk area to build, being well south of the major storm and seasonal cyclone-affected areas, I'm sure Amazon, considering their international scope, would have taken into account any reasonable level of risk when deciding where they want to put a DC to host their gear.
Australia has several 10s of terabits of international capacity, of which around 2Tb (from memory) is actually "lit". There's 4 main cable systems (AJC, SCCS, PPC-1 and SeaMeWe-3), a few smaller ones to surrounding nations (JASURAUS, Gondwana-1) and a handful of multi-terabit modern ones that are barely ticking over (like Telstra Endeavour).
The growth in capacity has drawn quite a few international service providers and carriers to Aussie shores, and the resulting demand for domestic capacity has done nothing but good things for the price and availability of rack real estate and domestic transit. Our domestic providers are all pretty healthy, just waiting to see how the NBN pans out.
I can't see any problems with a big cloud provider like Amazon entering the market here. If it doesn't start forcing storage and bandwidth costs down further I'll be quite surprised.
Slashdot Mirror
Then you're not paying attention; some already have.
Agreed. In every discussion I've had with customers about IaaS and cloud, the security aspect has been the #1 topic of conversation brought up by the customer. Closely followed by performance.
Businesses of all sizes and industries are very interested in all this mess in the cloud space.
One of my first IT jobs was working under an ex-Army sergeant IT manager. He used to work on Prime mainframes for the ADF before going corporate, and had a lot of experience with project management on the high end of defense & government IT.
We all had an immense amount of respect for him, whether it was me (early twenties) to the older 40/50-odd techs. If you screwed up, you got chewed out. He did a pretty decent disappointed-dad routine which was somehow worse than being yelled at.
We respected him because we knew he gave a crap and kept the "Directors of Micro-Management" well and truly off our backs for the most part, made sure Technical Services was involved in major decisions as much as possible. We would get warned if something nasty was trickling down from the top. It fostered a healthy team spirit too, everyone looked out for everyone else, helped each other with problems and so on. If a customer was causing problems or making spurious complaints, our manager would be out there looking them in the eye and asking if they had a business case for being difficult. On the other hand, if he thought an issue was your fault, you would know in very clear detail the hows and whys.
I only left that job when the company was acquired by another much larger one and the TSM was made redundant - though that's a much more involved story.
Ever since then, I've sought out companies with a similar "feel" to work for, and tried to foster a similar tone when taking up management roles. I've been successful finding places to work, and I'd like to think I've done as good a job as I can for the rest.
What's the point of this TL;DR? The biggest thing I've noticed is attitude makes up most of your workplace perception. I've worked in crappy places since (though not for long). Keep your head and chipper helpful attitude, useful people notice and come to you for direction instead of your boss and the vibe spreads. Walk around with a martyr complex and no-one will want to help. A healthy team means even if it's busy, it's not hard, and you have a lot of potential extra hands that may be more skilled than you. This goes a long way towards getting a recalcitrant boss on your side too, if they have any kind of management skill at all.
Can't really help with bad management, if they don't recognise your value, leave. You'll find out if you've overestimated on that part pretty quickly.
Since we're talking business users, VBscript macros. I know there's some interoperability but it's far from perfect.
Lets think of some other show-stoppers:
All of these things I've had to deal with for new customers within the last 3 years. There's many more examples than this. Usually these customers have grown out of being supportable by tiny 1 or 2 person IT shops or they've recently been stung by trying to go for the cheapest, poorly designed and implemented option tabled by some buzzword-driven cowboys. The amazing thing is, 1-2 years down the track, staff are happier and they've spent much less on support.
Years ago, working in education IT in my teens and early twenties, I spent a lot of time pushing LTSP, the K12 distros, OOo and friends for smaller schools (older computers, perfect for thin clients) and kindergartens. There was a bit of resistance due to unfamiliarity in some situations but us IT guys were pretty switched on with a rounded enough skill set to build and maintain a solution that worked in that kind of environment.
Small companies don't have a large, dedicated, government-funded IT crew on tap, and they can't afford $160/hr every time someone loses access to a calendar. They're very fast-moving compared to educational IT with highly varied requirements.
In most cases, Office "Just Works". When it doesn't, the 18-year-old helpdesk guy can usually fix it. You don't need to waste the time of a senior resource, or bill the customer for his time. For most businesses, having less headaches is well worth the additional cost of licensing.
I would be stoked if I could design in FOSS components into the solutions we support. We're trialled various components many times in the past. But I can't be confident they will work for and with all the requirements, and most importantly, that I can get resolutions for support issues immediately. Even with a support contract, outside of a few huge OSS powerhouses (eg, RedHat), support responsiveness and available skill pool leaves a lot to be desired.
Really? Looks to me like it's gone down consistently over the years, looking at past pricing on products since around 2003. The biggest anomaly in that is the abolition of Small Business Server resulting in a 20% markup on SBS/SBS Premium functionality after June this year.
Per-core licensing of SQL 2012 is (a fair bit) cheaper over 25-odd users (depending on feature set), SQL Express is free and each new version contains even more of the add-ons you used to have to pay for, Windows 7 & 8 was and remains much cheaper than XP was this time from release, Office has come down a little in price at Pro edition and a lot at the lower end editions or singleton products, Visual Studio is a bit less ridiculous and even has a free version, Server 2012 Std licensing offers effectively a superset of Server 2008R2 Enterprise's functionality at a slight increase over 2008R2 Standard pricing (there is no equivalent to 2008R2 Std anymore).
I could go on a bit more. There's been a lot of pressure from high-quality open source or lower cost alternatives, so MS has been pushed to show a bit more value.
Big licensing upgrades are admittedly horrifying to customers, when they see a $30,000 line item and ask what it is, and we say, "Oh, just Office". It's at the point where hardware is so cheap it has trouble making 1/3rd of the cost of a project, and it gets worse the bigger the organisation.
Turning off the key loses brakes and steering, if it even turns the car off. I'm pretty sure it's a push button start.
I'm also guessing that if the computer is really malfunctioning enough to ignore input from the accelerator it might also be ignoring input from the gear shift and start button. If someone has an hours worth of time and the mental capacity to call in a police escort they've probably tried a few things before getting to that stage.
The best bit of the article is this:
A Renault technician had been on the phone with police throughout the chase trying to help but couldn't come up with a solution.
Snort and associated tools aren't too bad, and should run on most Linux/BSD-based custom firmwares if the hardware has enough juice. A Cisco ASA with an IDS module is less good, but servicable. You'd need to use ASDM for monitoring unless you want to buy a super expensive monitoring suite.
The main issue is the amount of processing power and RAM required, especially if you're pumping through a lot of traffic. I run pfSense in ESXi on a little HP Microserver as my router. Using default settings with Snort, pushing through ~20Mbit of WAN traffic makes the poor little box bulge at the seams. It can be tweaked and the rulesets rejigged or disabled, but for Snort it's a RAM vs CPU tradeoff, and disabling rules reduces the level of protection. There is no way a cheap consumer router could have the resources to do it properly.
You're looking at more the SonicWall (ugh)/WatchGuard/NetGear/etc SMB/UTM router space. None of them qualifies as anything other than "terribly painful to use" for IDS/IPS.
If someone is using an iPhone, at some point it was connected to iTunes to activate it (or it wouldn't be working). The vast majority of people I'm aware of - technical or not - who have an iPhone also use iTunes, even if it's only occasionally on the work PC or at a friend's house. I had elderly relatives of friends asking me if I'd tried iOS 6, and showing me all the cutesy new features before I'd gotten around to upgrading myself. My iPhone is a work phone so I tend not to jump on updates immediately.
iOS updates over the air since 5.0.1, with a big obvious blob on your settings button if you don't, and nags every time you plug it into an iTunes-capable computer.
Your ISP doesn't have a dedicated interface just for you. And Bob who's on EFM. And Jill on VSAT. And Jim's Mowing with its 2 aggregated DSL lines.
Most ISPs I've seen collect flow statistics from routers, which works no matter what tail is in use and collects many customers' statistics at once.
It can still be handy to use a speed test attached directly to the provider's core network to determine where the problem is - my home ISP has a few large files on a customer website you can download, and their customer support will get you to try both an internal and external test site.
If the internal test is slow, then there's a problem on the backhaul between you and the ISP. If it works but external sites are slow, there's either congestion on the edge (usually pretty obvious on a traceroute, and your ISP should be very aware of such problems) or issues beyond the ISP's control. Still, getting traceroutes of such things allows them to pull their transit and upstream providers into line.
It's fairly common to use NetFlows or similar protocols to measure per-IP bandwidth at either the inner or outer edges of the network (depending on whether it's billed over all traffic or just external traffic to the ISP). NetFlows has a few issues where it under- or over-reports the traffic used by a network flow due to insufficient detail supplied by the protocol.
In most cases where an LNS is in use (such as with any PPPoX DSL connection), they can report down-to-the-byte accurate statistics, but often don't - it's easier for an ISP to centralise everything on NetFlows, which works fine for any end-user tail.
We have had a few problems with upstream providers giving us huge bills for specific customers, and us going back with a handful of graphs and spreadsheets to prove that their stats showing 50Mbit/s usage on a 20Mbit/s DSL circuit is highly improbable within the confines of modern physics, our monitoring of interface stats showing the same packet count with a far smaller payload.
I've had a couple of interviews like this. It really is a fantastic thing.
Once of those interviews did in fact end up getting me a job by word of mouth back in my younger days, it was completely worth it. Fast forward to today where it *is* a big part of my job to do this for people - as a presales solutions architect.
I did a double take reading TFS for this article, I can't imagine thinking like that. If you're asked a reasonable job-related question in an interview, you answer it. If you can solve one problem for them, it's likely you can solve more. If they're opening interviews to bring in outside opinions, all you've wasted is time. Someone who doesn't want to hire you or pay you for your labour obviously isn't worth working for, and if they get an hour or 2 of free successful consulting out of you, well, they're losing out if they don't hire someone who can pick up the state of their business and nut out a workable solution in that amount of time. I'd go find somewhere better, and a minute shift in attitude might help find that dream job.
As a Canadian, I like the connection we have to Australia through being in the Commonwealth, and never saw why so many of you guys got bent out of shape over what is really just a figurehead...
They've tried a few times and failed. There's no huge push for becoming a republic, the opinion polls are pretty even, and the last referendum was 55% against on all motions. Most people I know are pretty ambivalent about it, not really worried either way.
Me personally, I don't believe changing the status quo benefits anyone but the politicians. We have a historical connection to Britain. Our parliamentary and legal authority is vested in our own institutions. I'm somewhat wary of letting our pollies re-write and amend big chunks of our constitution to allow our head of state to use a different title on the departmental letterhead.
I'll have the crab juice...
...which is a business model built around distributing FOSS.
The view that the kernel has of memory allocations is somewhat different to what the application sees - most of the work is done in the libc, which directs the kernel to map regions and size the heaps that it malloc()'s from.
Usually, you'd have GC handled by other libraries, for instance the Boehm-Weiser GC. If you're using C++, Boost also has a few wrappers and implementations of garbage collection algorithms for a variety of use cases.
From memory I believe the kernel CMA is mostly related to in-kernel allocations, so drivers and kernel threads.
It sounds like a peering arrangement turning into transit. There was a comment earlier about extra detail from French language reports on this story; Google is trying to break into certain African markets that Orange have a large presence in. Orange, like many large carriers, is having a hissy fit over the traffic volume coming from Google services. This would be an olive branch from Google being used as a back-scratcher.
It is not uncommon for Google to co-lo equipment and caches in the datacentres of larger national carriers. Usually this is peered directly or hosted by a common peering IX. Google in this case may have agreed to pay directly for transit.
It's difficult to tell exactly what is going on without more info, so far we have a bragging CEO and no real solid detail.
I'm wondering if there'll be any announcements about the Gertboard and similar prototyping tools for the Pi? It's a shame it went off sale about a week before I received my Pi.
Consider:
Striking: a combination of denying your services (and only yours) to an employer with (usually) highly visible protests and PR to get the reasons for the strike out to the general public, thus gaining their support and pressuring politicians. In most locations, it is illegal to actively interfere with the employers' ability to do business beyond withdrawing your own labour and expertise - blocking access to sites, harassing customers, etc will probably attract police attention. If a group of other people decide to do the same, they've done it under their own steam and should not be coerced to do so.
DDoS: actively suppressing the target's ability to communicate and do business with very little cost or effort, and a high potential for serious collateral damage to the operations of completely unrelated individuals and businesses along the way. In some rare instances individual participants may volunteer their resources to the DDoS, but much more often a lot of traffic is coming from compromised systems. DDoS and systems intrusion are both already criminalised actions.
I'm 100% for the ability to protest or strike - I've done my share in the past. I'm 100% against bored kids getting their mates together to smash small websites into gravel "for the lulz".
You can argue that it'll only be used as a responsible form of protest against those that deserve it, at which time my eyebrows will climb my forehead and burrow under my hairline. The amount of effort required to "protest" is much too small, the relative damage against small operations and individuals far out of proportion. If we would like to see spam outfits "protesting" against Spamhaus' DNSBL resolvers, pro-X advocates "protesting" against the blogs of their pro-Y opponents, kids targetting the local donut shop because they're bored and want to give them a huge bandwidth bill, sure, lets allow legal DDoS. Perhaps we could allow governments to set rules on what constitutes "fair protest" to solve this problem? I'm not really warming to that idea either.
they all use the same libraries
They all use slightly different versions of libraries, each of which introduces slightly different bugs and issues into the environment.
As long as you have the right dependencies on the package, its all gravy. packages are just renamed compressed archives, and they can easily be repackaged in 20 lines of bash, something that runs on all linux machines.
You still need a deployment build environment and a test platform for every target - which would include every sub-distro you want to support, as they all have different package sets right down to libc, every "branch" (testing, stable, etc) and so on.
It's quite difficult to package games using the native package manager in any distro and expect the package to have any kind of supportability or longevity. If the vendor has to support it over time, they're only going to support a select few distros, or they'll be enjoying stress headaches until support is withdrawn.
The easiest options for binary software distribution are what we've seen many times before: statically link the whole lot (or at least remove external dependancies) and distribute a tarball (like id did with all its Linux releases), or if you want to get fancy, an installer script/GUI like VMware or WordPerfect (back in the day). This can be wrapped up in a native package but it's not going to be adding much value beyond showing up in the programs list.
SBS is discontinued past SBS2011 - you just have to buy the separate products and pay for them unbundled (about a 15% increase in price on a project, depending on your CAL model and quantity - going on a 20 seat basis here). Server Essentials 2012 is something else that's been around for a while with one name or another.
You're entirely correct about them pushing the cloud model with these changes - we've had a bit of pressure coming down the channel to push Azure and 365, which has been met on our end with a mixture of quiet sniggering and outright mirth.
OK, I'll bite.
A big part of my job (at an IT integrator) is going out with salespeople to chat with customers when potential projects are looming, determining requirements and a suitable solution. From that perspective, we've had a lot of questions asked about Windows 8, so we've run some demo units up to loan out for the curious. Our help desk also had a couple of their workstations converted to familiarise themselves with the environment.
Within a business environment, the feedback has been entirely negative. We have yet to do a Win8 rollout within a client. Internally, those deskbound staff who got a Win8 PC have since reverted to Windows 7 after trying really hard for a week or 2 - I'm not sure why, it more likely relates to us having a lot more clients running Win7 and helpdesk staff wanted to have an environment to match. There is still a spare Win8 demo PC for internal use by any tech team members. Even reverting to a "classic Aero" interface with a start menu was frustrating.
Our customers found the interface very confusing (far more so than WinXP to Vista). Even with classic mode, they did not like the distractions, the interfaces, and felt that nothing in Windows 8 added business value beyond what was available in Windows 7.
All of that is pretty obvious - I never expected something touted as effectively "Windows Touch Screen Edition" to have a huge business focus. However, I've been talking to our various IT contacts at customers - we would talk to a few hundred at least once a month, managed services and whatnot - curious about Win8 for their own use on laptops, tablets and home PCs. I gave them my opinion of it (which isn't high from my admittedly limited exposure). The feedback there has been almost entirely negative. Even customers who adore their new Windows Mobile devices hate the similar interface in Win8. I don't think we've sold a single PC without downgrading to Win7 first.
Very few of these people have used anything but Windows since '00. We HAVE had a big spike on MacBook purchases recently.
Admittedly, we're an integrator and we do not box drop or sell retail product. Harvey Norman might have a different story, but they don't have to deal with the support issues on the back end. We get that when a CFO calls us because his kids can't use the new home PC - and we're not going to tell him we can't help.
Consider that a lot of central phone exchanges and DCs are in the Brisbane CBD, and most kept working in early 2011 when a good chunk of the state's land area was underwater: http://forums.whirlpool.net.au/archive/1616757. The locations I've got gear (PIPE DC3, Pegasus and Fujitsu) were well out of the flood zone, being located in areas above any reasonable flood level - if they had gone underwater, there wouldn't be a Brisbane anymore.
The main drivers of DC location are cost and utility. Cost includes the normal stuff like power, real estate, connectivity and so on, but also factors in risk, redundancy, insurance and mitigation of each of these factors. Even though Sydney is a relatively low risk area to build, being well south of the major storm and seasonal cyclone-affected areas, I'm sure Amazon, considering their international scope, would have taken into account any reasonable level of risk when deciding where they want to put a DC to host their gear.
If you're worried about refreshments, see if you can convince Amazon to move the DC to Melbourne.
Australia has several 10s of terabits of international capacity, of which around 2Tb (from memory) is actually "lit". There's 4 main cable systems (AJC, SCCS, PPC-1 and SeaMeWe-3), a few smaller ones to surrounding nations (JASURAUS, Gondwana-1) and a handful of multi-terabit modern ones that are barely ticking over (like Telstra Endeavour).
The growth in capacity has drawn quite a few international service providers and carriers to Aussie shores, and the resulting demand for domestic capacity has done nothing but good things for the price and availability of rack real estate and domestic transit. Our domestic providers are all pretty healthy, just waiting to see how the NBN pans out.
I can't see any problems with a big cloud provider like Amazon entering the market here. If it doesn't start forcing storage and bandwidth costs down further I'll be quite surprised.
In comparison, it's quite a bit more difficult. But yes, there are complex tools available to handle all the nasty stuff behind the curtain.