IPv6 end-users represent >30% of the traffic hitting some of the services I operate.
711 karma · joined October 5, 2014
When a negotiation fails, it's not unusual for the purchaser (or employer in this case) to be more upset than the seller.
If she's a quality employee that just graduated college, it stands to reason she'll be able to find more gainful employment...
But one has to wonder if this is just an elaborate troll.
When I say issues, I mean "something, somewhere in the world is broken". More POPs and more peers means more outages. This is probably especially true for their POPs in less developed regions of the world.
Cloudflare also has a team monitoring for issues and it's usually pretty easy for them to work around the issues that come up. Their design makes it so that even the loss of an entire POP isn't a big deal.
As their service continues to get better and more reliable, their status page will actually look like it's getting worse and worse. But that's because they are being open and transparent about day-to-day operational details.
Triangular routing is a use case which becomes lost. As does the the ability to easily set up certain kinds of peering/multihoming connectivity.
The larger and more complex a network gets, the harder it can become to determine if a source IP is "correct".
... only I had already spun up a new server. Because the existing instance wasn't working. Getting their system to terminate the old instance required several attempts and force termination before it succeeded.
Considering their scale, I would expect better handling of the edge cases.
I agree it's not ideal to have filesystems do this, but it also simplifies a lot of engineering. And I think direct user exposure to a filesystem with a POSIX-like interface is a paradigm mostly on the way out anyway, meaning it's increasingly feasible to design systems to not exceed a safe utilization threshold.
Agreed
>> It's not as if the user interface for a router's configuration is bleeding edge technology.
Not sure I agree. Have you seen some of these modern interfaces? They are doing deep packet inspection, providing many different visualizations of usage, geo-location, captive portal, ... the feature list - and attack surface - is huge.
(2) If your pool is big enough/expensive enough, surely you've also budgeted for backups.
Doing so requires that you have enough zfs filesystems in your pool (or enough independent pools) that you have the free space to temporarily have two copies of the filesystem.
It's also a simple and state-free way of influencing other devices on the network. Arbor is an interesting example - when doing DDoS mitigation, an approach which doesn't require a lot of resources or statefulness is extremely useful.
On one hand, I want to agree with you: this is terrible and makes everyone else's life harder.
On the other hand, maybe its TCP's resource requirements that are at fault.
Manufacturers can get their drivers merged into the mainline kernel and they won't have a driver backward compatibility issue to worry about.
They choose not to for business reasons (market segmentation/artificial hardware limits via drivers) or because they are being cheap. And because much of the market doesn't know enough to care, it mostly works for them.
In the long run I think it limits the IoT ecosystem from progressing and it's a practice which must stop. If your hardware is not fully supported by the mainline kernel, it may as well not exist.
For single family detached, the value of the building is not a big deal. It's the dirt the building sits on that people are buying. Prices went up ~30% over the last year in Vancouver.
There are also buyers who think nothing of tearing down a relatively new building because it's not quite what they want.
As long as the city remains a place people wish to live, the dirt will retain value. Perhaps it goes down 50-80% - you can live on your land or you can use its cash flows to help cover your living expenses. As a safe-haven, it's not a bad option.
The price in CAD/USD/EUR/CNY/XBT is almost irrelevant to these owners.
The problematic purchasers in my view are the locals. Many are buying because the market is going up - "I'd better buy now or I'll never be able to afford it". In a falling market, that demand many not exist.
Though I also remain unconvinced that the foreign demand will remain strong in a falling market.
What's going on in Vancouver isn't sustainable in any traditional sense. But the market can also remain irrational longer than many can remain solvent.
I agree that the phenomenon of properties remaining vacant represents a failure. I'm curious what opportunities you think there are. I suspect that many of the opportunities would require regulatory changes to be viable, but I'm open to a different perspective.
Regarding Chinese people understanding capitalism, I suspect there's a very different view culturally. I'm told that the advertising for financial products is quite different - in North America we tend to expect a convervative branding. In China you'll see investments with more of a WeChat approach (including being actually promoted on WeChat). Property rights were recently codified in China - i.e. 2007 or 2008. Real estate is mostly only available as leasehold in China. There's a lot of people who recently became quite wealthy quite quickly. I think it's very likely that they have very different understandings of capitalism than North Americans, who also have a different understanding than Europeans.
A 10-20 year old building in Vancouver will have many units with kitchens that have literally never been used. There's wear and tear and a lot of extra work and risk that comes from renting a place. And as odd as it may sound, the resale value can take a hit when a property is "used".
Unlike San Francisco, rents in Vancouver have not gone through the roof. Recently AirBNB seems to have put some upward pressure on rents, but rents still remain fairly detached from capital costs.
Frankly, the rental income is almost a rounding error. Vancouver real estate is about capital, not cash flow. It's a place to store capital (which in and of itself is quite valuable for many foreign owners), and current supply/demand trends also lead to price appreciation.
Traditional views of real estate market fundamentals are nearly irrelevant to the observed behaviour of the Vancouver real estate market.
I wish excluded keywords were the only frustration. Their index seems to have gotten considerably worse at dealing with any somewhat "long-tail" query. I'm not sure why.
I suspect they've won the spam battle by aggressively purging from their index, but in doing so they've lost the spam war because the false positive rate on their approach is way too high. Or maybe I'm completely off base.
It taunts me and tells me that I'm sorry Daniel, I've decided to exclude the most important keyword from your query.
Domain-specific search tools - e.g. for programming, Dash.app - have become an increased part of my workflow partially because Google simply isn't giving me the results I am looking for.
But when it doesn't, debugging might actually be simpler with less black boxes between you and the metal.
GPLv2 was fine. But GPLv3 can make one look for alternatives. AGPL will cause many to simply lose all interest. I suspect GPLv3 is why we see so much MIT/BSD/Apache licensed software these days.
Furthermore, healthcare costs don't reflect the complete picture. Firstly, lost productivity is an issue. Secondly, we may not want a society where ritual suicide is encouraged at a pre-determined age with the goal of reducing healthcare costs.
If the IP address of the microservice nginx depends on changes, in a standard nginx config, nginx will not be updated. A Kubernetes health check could also hide this issue.
IP addresses of upstream servers change, and sometimes DNS doesn't resolve until another service is brought up. Nginx does not work well in such an environment unless you (1) have ugly configuration hacks, (2) add OpenResty/LUA to your configuration, (3) add a watchdog process, or (4) buy their nginx Plus version and keep track of how many instances of nginx you have.
I want simple. If you need a webserver that can use DNS, Nginx is a lot of extra work.
A couple days ago I contributed https://github.com/DanielDent/git-annex-remote-rclone to git-annex, so in theory you could use Dropbox as a backend for git-annex if you wanted to (I haven't personally tried it with Dropbox).
But if CloudFlare were to shut them down without due process, that would damage their credibility - to the point where I would feel uncomfortable doing business with them.
I'm sure there's a group of people where CloudFlare's stance is damaging to their reputation.
But there's also a group of people where any other stance would damage CloudFlare's reputation.
Think also from the customer standpoint: people who need DDoS protection are people who are having their site shutdown without due process (by criminals who launch DoS attacks). When the company you hire to fix the problem becomes part of the problem, it's not good - and they would be part of the problem if they offered a 'send us an email and we shut down our users' denial of service attack vector.
It's surprising to hear this. I would have thought that your backup workloads would leave enough unused IOPS that you wouldn't be anywhere close to hitting a performance ceiling.
What metrics do you optimize towards? Tail latency?