Slashdot Mirror
Airlines Suffer Worldwide Delays After Global Booking System Fails (bloomberg.com)
rastos1 writes: Airlines worldwide were forced to delay flights Thursday as a global flight-bookings system operated by Amadeus IT Group SA suffered what the company called a "network issue." Major carriers including British Airways, Deutsche Lufthansa AG, Cathay Pacific Airways and Qantas Airways were among those reportedly impacted by the outage. Singapore's Changi airport said via Twitter that a technical issue affecting some operators was delaying the check-in process, with boarding passes having to be issued manually. "Amadeus confirms that, during the morning, we experienced a network issue that caused disruption to some of our systems," the Madrid-based company said in a statement. Technical teams took immediate action to identify the cause of the issue and services are "gradually being restored," it said.
I'm happy the slashdot IT team found new jobs so quickly.
"National Security is the chief cause of national insecurity." - Celine's First Law
Since this is critical software, rewriting it in a modern language designed to ensure security and reliability may be a good idea. The obvious candidate is the Rust programming language. It is being created by Mozilla for use with Servo, their next generation browser engine. Rust is being built from the ground up to be safe and secure and reliable. This is just what's needed when creating complex software systems that have to work properly all of the time, like airline booking systems.
Your words are in English, but your post makes no sense.
That would have been 17 Moore's Law generations ago! In human terms, it like looking at the farming methods or weaving techniques or marine navigation procedures or military maneuvers of 1592!
sed -e 's/Chuck Norris/Rajnikant/g' joke > fact
First it was Slashdot, then it was the airlines.
As everyone is surely aware, /. and SourceForge have suffered some serious outages and problems over the last two days. Reportedly it was due to equipment failure. The big question is, would hosting /. in cloud infrastructure like Amazon's AWS or Microsoft's Azure have avoided these outages? The answer could very well be a resounding "YES".
Modern cloud platforms allow for an unparalleled amount of redundancy and geographic distribution. Web sites are no longer constrained to just a single data center, but instead can be distributed across the globe with ease. The use of virtual machines and high-level services makes hardware more and more irrelevant. In fact, such cloud platforms are typically designed with hardware failure in mind, with such failure being a routine part of their massive-scale operations.
I think that the recent outages here highlight the need for web sites and other online services to take advantage of the expertise and efficiencies offered by the leading cloud platform providers. A web site in 2017 shouldn't just run from a server or two in a single data center. A web site in 2017 should be globally-distributed, with redundant web servers across the globe. The best part of this is that thanks to economies of scale, cloud-based hosting can often be significantly cheaper than traditional dedicated hosting.
If the whole system is this fragile, it will be more cost-effective to select a stronger platform and development tools, and begin redesigning it now.
I hear that ADA works very well for building reliable software that doesn't exhibit surprises or unexpected behavior.
By the second paragraph it becomes clear this is just an advertisement in disguise. But thanks Anonymous Advertiser!
Are these airline information systems really all that fragile in this sector? I know we all say that, but I personally don't have a F clue; I'm 100% media driven on this from what I read, consume or read-between-the-lines. I'm hoping someone close to this could chime in or reply...
With any one of us with any moderate amount of IT experience in the trenches and at any level that's support any ops or for-profit system, It's hard to dismiss a generic statement such as network issue. I know management I've worked under in the past at other organizations, private and government, would pre-can some huggable and down-played message like that --- and I totally get it; it's embarrassing on any level for any end-user disruption, but we'll never know why.
With the amount of breaches, DDOS's and what seems like this popular resurrection of using the word 'Hacking' like we are all hoping a Hackers reunion happens with Jonny Lee Miller and Angelina Jolie is just nauseating, but a very true reality anymore with the lack of implementation over security practices.
AWS has had plenty of outages.
Personally I don't think there is any such thing as "technical issue". There are resourcing, risk management, and personnel management issues. I've built systems with the right teams before that could stand anything short of a nuke, and we came in under budget. Honesty and the right people give you results.
...The best part of this is that thanks to economies of scale, cloud-based hosting can often be significantly cheaper than traditional dedicated hosting.
Actually, the best part of this will be listening to the grand excuses justifying "cheaper" after the cloud provider is hacked.
Targeting and hacking a single organization is one thing. Hacking a cloud provider can result in exponentially larger gains, so it tends to make the (very few) major cloud providers huge fucking targets. And for the "will never happen" crowd, let me guess...you were an Equifax customer too? Or perhaps you used a bitcoin exchange...
they don't have local console and you are not able to use your own ISO to install an OS.
Holy fuck. The idiocy that we see here at /. these days is astounding! Have you ever actually used any sort of a cloud provider?
ISOs are a relic of the 1990s. The recommended best practice today is to upload a preconfigured VM image.
Here are some instructions for doing it using Microsoft's Azure platform.
And here is some information about doing it using Amazon's AWS platform.
The advantages of this approach should be obvious. But since you seem oblivious, let's look at some of them. First of all, you're not stuck using an ISO image as the installation medium. Many Linux distros and even other OSes offer network-aware installations that bring in only the software you actually need. So you can build your VM using only the software you want, and then upload it to your cloud provider. There are other benefits, like being able to configure the VM locally, rather than remotely. You inherently get a local initial backup. It's often quicker to upload a small 100 MB compressed VM image than it is to upload a 800 MB or even 4 GB ISO image.
And a local console is irrelevant in the world of virtual machines and virtual storage. If you have a problem with a VM, you can disconnect its virtual disk, attach the virtual disk to another working VM, apply whatever fixes are necessary, and then reattach the virtual disk to the initial VM. In the extraordinarily rare case that there's a problem with the VM, then you just spin up a new one and destroy the old one, after detaching any attached virtual disks to preserve the data.
You're literally stuck in the 1990s, from what I can see. The world has moved on long ago, but you're still dicking around with primitive approaches. The problems you're facing don't even exist for the rest of us because we moved past them over a decade ago!
http://diehard.wikia.com/wiki/...
Requiem for the American Dream
The core software is written in C++, and originally very well too I might add. However it is HUGE, and no one person or even one team understands even an overview of how it all fits together, never mind the specifics. Which is fine until there's a cascade failure and problems get batted around from one team to another. Also because no one really understands the low level side when new features were added someone would just bolt more code onto the top of side instead of finding out if the core stuff could already do it and after years of that you end up with something immense, unwieldy and essentially ineffable in its whole to even top class programmers.
Posting A/C for obvious reasons.
Requiem for the American Dream
... but pushing your pet language in every goddam comments section is a perfect way to make people get sick of hearing about it and give it the finger before they've even tried it. Who knows, perhaps thats your intention. Either way, give it a rest you buffoon.
And it doesn't matter what language they're written in , the fragility generally isn't down to a low level language issue such as memory, threading or pointer issues (though obviously those errors happen too), its usually a logic problem in handling edge cases, unexpected code paths and errors correctly. No language is going to save you from broken logic however much their proponents would pretend otherwise.
Too bad they're not hiring "miracle workers," eh creimer? Actually, given the availability of the site, I dare say you'd fit right in in their IT department!
Every human built system can suffer from technical issues. Saying otherwise is just pretending the problem doesn't exist.
" I've built systems with the right teams before that could stand anything short of a nuke"
You're modest arn't you. Systems always look bullet proof - until they go wrong. I doubt yours are any better or worse than hundreds of others that have been written to be resilient.
Surely you can provide a link to such a job posting, creimer? You're not just endlessly repeating the same meme because you think it's amusing?
Too bad they're not hiring "miracle workers," eh creimer? Actually, given the availability of the site, I dare say you'd fit right in in their IT department!
Who the fuck is creimer??
Surely you can provide a link to such a job posting, creimer? You're not just endlessly repeating the same meme because you think it's amusing?
Give it a rest, FakeFuck39, and get a fucking life.
Airlines Suffer Worldwide Delays After Global OVERBooking System Fails
I've built databases that work flawlessly for a year and stop working when I'm on vacation without explanation. Shit happens sometimes.
"Action without philosophy is a lethal weapon; philosophy without action is worthless."
wait, say what?! Slashdot was down?!
I've been troubleshooting my network for the last 3 days!
So that's a "no, I can't provide an example of a job posting that does this"?
tsk tsk, creimer.
...is mostly syntax-equivalent with Oracle PL/SQL. The GCC toolchain targets ADA with GNAT. As such, it would obviously link against C.
ADA is quite old and is likely missing many of the features you've outlined. Some of them may be present in the popular descendant of ADA known as SPARK.
It is well-known that our software breaks far too much. Denying the problem does not solve it. ADA was designed to address this issue head-on, which is why Boeing's airplane control software is not written in C.
You are, that's who.
From a person who has had similar international headlines for systems that I can impact. My heart goes out to you. System failures are never fun, failures that affect a lot of customers are just plain stressful. Document processes and learn from this event all the you can. Customers care most what was learned and how to prevent this and future scoped events from occurring again.
Hang in there!
There is or can be built a machine that can simulate any physical object. -Church-Turing principle
If it is happening at 6:45 in the morning just ask the cleaning staff where are they plugging in the vacuum and the coffee maker.
Sadly, for contract reasons I can't give more details than the published one.
https://www.dallasnews.com/business/southwest-airlines/2016/07/30/southwest-ceo-router-failure-grounded-flights-equated-thousand-year-flood (paywalled but you can read it if you have NoScript ore equivalent installed)
https://www.wsj.com/articles/why-a-single-failed-router-can-ground-a-thousand-flights-1489743001 (paywalled)
The airline business suffers from not only aging systems, but also from highly interconnected, interdependent and extremely complex systems that keep growing in complexity and new systems and companies get connected all the time.
Many of these systems are mission critical. The amazing thing is that all this is working almost ALL the time. The bad thing is an error on a critical path leaves everybody grounded.
I have to acknowledge that Amadeus systems are much more stable than others in the sector (saber had a few high severity failures in the last couple of years, like this one https://www.dallasnews.com/business/southwest-airlines/2016/07/30/southwest-ceo-router-failure-grounded-flights-equated-thousand-year-flood )
In Amadeus there are some nice things going on like it is published here:
https://www.thecompanydime.com/mainframe/
and here:
https://www.nextplatform.com/2015/08/04/amadeus-takes-off-with-containers-and-clouds/
Even if not everything is yet "in the cloud"
Besides that, and here my rant about the company:
I believe the main problem is incompetent management and too much politics in the middle that play to get everybody not unhappy enough, instead of doing what needs to be done. This plus HR that are mainly there to make senior management happy instead of doing the right thing doesn't add to the equation. This leads to: most of the really good ones either quit or stop caring about (although some manage to get to do really well what they want to do), the mediocre ones getting to most of the work and the bad ones that like talking and taking credit for other's people get to manage. Amadeus deals with developers and engineers only as a "cost" and not as an investment (even if their words say the contrary).
This leads to slow change and errors in technical and strategical decisions. I know of some, because I've fought for some while I still cared about.
What seems to have happened (in my opinion) can be dealt with (and I'm sure they will because some heads might be rolling right now) in the future.
So that's a "no, I can't provide an example of a job posting that does this"?
tsk tsk, creimer.
Who the fuck is creimer??
You are, that's who.
Anyone who disagrees with you is creimer. Gotcha. Have you ever thought about getting psychological help?
Reportedly it was due to equipment failure
If you trust some with the name "Logan, A Bot"...
Ezekiel 23:20
Who the fuck is creimer??
Isn't that the guy living next door to Alice?
Ezekiel 23:20
He's using the wrong words, but essentially he's saying that they get to charge you for their mistakes caused by having lousy IT that is lousy because they are saving money by not spending it on training.
Creimer is a sad troll, a symptom, but not the proximate cause, of Slashdot's death spiral.
The disadvantages of using a potentially hacked OS provided by someone who definitely cannot be trusted, should be obvious.