Today, with our network nearing 80 global locations, we’re pleased to announce support for WebSockets for all our customers, including Enterprise, Business, Pro, and Free, with resources allocated by plan level.
What are WebSockets?
If you don’t want to read RFC 6455, then this short paragraph from our previous blog post explains:
The WebSocket protocol is a distinct TCP-based protocol, however, it’s initiated by an HTTP request which is then “upgraded” to create a persistent connection between the browser and the server. A WebSocket connection is bidirectional: the server can send data to the browser without the browser having to explicitly ask for it. This makes things like multiplayer games, chat, and other services that require real-time exchange of information possible over a standard web protocol
There’s a lot more technical history in that post covering how we modified NGINX to support a huge number of connections through port reuse. But the bottom line is that WebSockets are a vital technology for web sites that need to get real-time updates for features like social feeds, multimedia chat, collaborative editing, multi-player gaming, stock updates, and more.
Performance & Latency
Because the WebSocket protocol creates a single connection and doesn’t require HTTP requests, WebSockets can provide anywhere from 500:1 to a 1000:1 reduction in unnecessary HTTP header traffic, compared to HTTP polling solutions.
The server no longer needs to wait for a request to come back before it sends new data; it can simply push the data to the client the moment it has new information. Tests have shown a 3:1 reduction in latency compared to other polling solutions.
Modern Internet, for More People
Last week, we expanded our deep support for HTTP/2 to include Server Push. HTTP/2 was finalized as a spec in 2015, so almost a newborn child as technologies go. (We’re excited to help HTTP/2 grow up fast.)
By comparison, WebSockets is already a toddler, as RFC 6455 was finalized in 2011.
But, perhaps because of complexity or expense, WebSockets hasn’t seen as wide adoption as we’d expect. CloudFlare is removing those obstacles, and frankly we’re curious to see what happens.
We are building a better Internet, and while that may sound like a platitude… making modern tech easy to use matters. We’re eager to see what our customers do to play, grow, and thrive with another tool available.
In addition to CloudFlare’s core Performance and Security capabilities, CloudFlare’s network service also includes Websockets support for all customers.
CloudFlare will accelerate your WebSockets app with a large global content delivery network, enhance reliability with DDoS protection and increase security with our IP and Web application firewall. Check out our plans page for more information.
WebSockets represents a giant leap for real-time web applications with features like social feeds, multimedia chat, collaborative editing, multi-player gaming, stock updates, and more.
Standardized by the IETF as RFC 6455 in 2011, WebSockets is a protocol that provides full-duplex, bi-directional communications over a single TCP port.
Because WebSockets creates a single connection and doesn’t need multiple HTTP headers, WebSockets can provide anywhere from 500:1 to a 1000:1 reduction in unnecessary HTTP header traffic compared to HTTP polling solutions.
The server no longer needs to wait for a request to come back before it sends new data; it can simply push the data to the client the moment it has new information. Tests have shown a 3:1 reduction in latency compared to polling solutions.
Protecting your mission critical real-time WebSocket applications
The ability to protect and accelerate WebSockets has been one of the most requested features from our customers.
CloudFlare offers real-time protection against attacks vulnerabilities including cross site scripting attacks, SQL injections, comment spam, excessive bot crawling, to email harvesters, and more. Learn more ›
We’ve built the next-generation global content delivery network that distributes your content around the world. Our proprietary software allows us to process millions of requests per second efficiently. Learn more ›
Modern denial-of-service attacks present complex and overwhelming security challenges. CloudFlare is provisioned at the network edge matching the sophistication and scale of such threats. Learn more ›
CloudFlare makes more than 2,000,000 web properties faster and safer. Join
At WWDC 2015 we announced the transition to IPv6-only network services in iOS 9. Starting June 1, 2016 all apps submitted to the App Store must support IPv6-only networking. Most apps will not require any changes because IPv6 is already supported by NSURLSession and CFNetwork APIs.
If your app uses IPv4-specific APIs or hard-coded IP addresses, you will need to make some changes. Learn how to ensure compatibility by reading Supporting IPv6 DNS64/NAT64 Networks
Original URL: http://feedproxy.google.com/~r/feedsapi/BwPx/~3/Y2yfsCm_Pgw/
There are a lot of different ways to advocate for a universal basic income: as a socialist-ish equalization of resources, as a libertarian-ish replacement for the welfare state, as techno-utopian social engineering for the coming robot-driven end of work. In his latest Janus Investment Outlook, Bill Gross is a little techno-utopian, but he also argues for the universal basic income as macroeconomic policy: helicopter-money UBI.
Money for free! Well not exactly. The Piper that has to be paid will likely be paid for in the form of higher inflation, but that of course is what the central banks claim they want. What they don’t want is to be messed with and to become a government agency by proxy, but that may just be the price they will pay for a civilized society that is quickly becoming less civilized due to robotization. There is a rude end to flying helicopters, but the alternative is an immediate visit to austerity rehab and an extended recession. I suspect politicians and central bankers will choose to fly, instead of die.
Beyond the socialist/libertarian/tech/macro cases for UBI, its deep appeal is the possibility of redefining human worth and dignity without reference to work. People in modern capitalist economies are expected to work, and their self-conception is bound up in the job they do and how good they are at it. In a post-scarcity world where the robots do the labor, how will we fill our time? How will we find meaning in life? Gross has his own ideas, which apparently involve drum circles:
How to live a life – this Shakespearian brief candle? Should I listen to the beat of a bass drum instead of an ancient tom-tom? Would I dare dance to strange new music with a different step? “Forward” is my futile response. Forward – with difficult questions. John Denver expressed it succinctly, “If there’s an answer, it’s just that it’s just that way”.
Imagine a young Bill Gross, offered a basic income, free of the constraints of needing to earn a living. Would he still have become an obsessive bond manager? Yes of course he would have, come on. Gross has been open about the fact that he’s not in bond investing for the money; he’s in it for the fame. And there is no universal basic income of fame, though I guess Twitter is getting us pretty close.
Also he is so competitive. He sold some fancy stamps the other day and donated the proceeds to the Pimco Foundation, presumably in part so he could throw this shade at his old employer:
“I have a special affection for the Pimco Foundation, which I co-founded in 2000 and 100 percent funded for its first two years,” Gross said in the statement. “Irrespective of my current employment status, I am still a firm believer in the Pimco Foundation’s mission to help people around the world reach their full potential by engaging, empowering and investing in communities.”
One really appealing potential blockchain application is the idea of “smart contracts,” where you program some series of cash flows and asset transfers to self-execute based on events in the outside world. So a smart call option automatically routes a share of stock to the buyer, and cash to the writer, if the stock is above the strike price at expiry, and not if it isn’t. Or a smart interest rate swap automatically calculates and sends payments between the parties. Everything moves on the blockchain; no one needs to pay attention, or devote time to settlement processes, or keep an eye on counterparty credit. The contract just handles all of that itself. It is smart.
But in the actual world there are only so many contracts that pay off solely based on easily verifiable external events. Lots of weird stuff happens in the actual world! I used to negotiate corporate equity derivatives, and the contracts would have like a page or two of economic terms expressed in a math-y shorthand that you could easily program into a smart contract. Then there were like ten pages of “well if this general category of weird thing happens then we’ll adjust in a way we consider fair” sorts of terms. The world is complicated, and it is almost never sensible to write a contract that considers every possible future contingency. (This is called “incomplete contracting.”) Some stuff is left to trust and goodwill and after-the-fact negotiation between the parties, or, failing that, to the courts to figure out.
Introducing trust, subjectivity, and national court systems into smart contracts dims their luster a bit. But there are alternatives, maybe? Here is a fascinating Izabella Kaminska post on “Decentralised courts and blockchains,” quoting Vitalik Buterin of Ethereum:
One crypto-institution that would be very useful for a large set of different applications is a mechanism by which a user could ask a question, expressed in the form of English text, and have a decentralized mechanism, perhaps based on schellingcoin, Martin Koppelmann’s ultimate oracle, subjectivocracy (a very similar concept to slock’s DAO splitting) or some other scheme with similar properties determine the answer, and then send a callback and a log to the user who asked the question. To achieve scalability, a multi-stage scheme where only a few randomly selected judges look at each question by default, and are incentivized by the threat of a larger “supreme court” contradicting them, is probably optimal.
Sure, why not. If you can get people together to trust each other, and the system, enough to build a currency, why not have them build a court system too? Kaminska is skeptical about enforcement:
Dispute resolution without such support would take on an entirely different and unappealing course. The threat would have to come from elsewhere. Most likely, in the form of a devolution to a Wild West framework, complete with bounty hunters, mercenaries, guns, the threat of being run out of town or shunned, death threats and on-the-spot Judge Dredd-style judgments.
But I don’t think that’s the right objection. The idea, as I understand it, is still to build a smart contract, one that is self-executing based on external events except in a few weird cases. In those weird cases, the smart contract appeals to the Court of Random Judges, and the court decides, and it sends its decision to the contract with whatevercoins, and then the smart contract still self-executes. You don’t need a bounty hunter; the call option just sends you your stock automatically.
People are worried about Deutsche Bank’s ability to pay coupons on its additional tier 1 capital securities, but this should … um … make them feel … worse?
“What we do not disclose, and can’t disclose, are additional reserves” that may come from moving profits at subsidiaries or re-valuing assets on Deutsche Bank’s balance sheet, CFO Marcus Schenck said in a conference call on Wednesday. “Let me call it hidden reserves,” he said in response to an analyst’s request for clarity on what would be available in a crisis.
A bank is a collection of probability distributions: You have a bunch of assets, and each asset is worth something between $X and $Z, probably somewhere around $Y. Management has plenty of flexibility in deciding how to value each of those assets. When things are fine, you maybe shade a bit closer to the lower valuation, to build yourself some cushion for later on. When things are bad, you maybe shade a bit to the higher valuation, to manufacture some profit to offset your losses. All of this is both inevitable and frowned upon. You are not supposed to say it on conference calls! Come on.
Elsewhere in banking news: “Through layoffs, turnover, and moving employees to new ventures,” State Street “aims to become a more tech-driven company less reliant on faxes and manual trades.” Faxes! The blockchain courts can’t come soon enough. Also: “Credit Suisse Said to Cut 180 London Jobs, Mostly in Trading.” “Barclays Plc said it sold about one-fifth of its stake in Barclays Africa Group Ltd. for about 13.1 billion rand ($879 million) as part of Chief Executive Officer Jes Staley’s plan to overhaul the British lender.” And: “JPMorgan Securities on Hong Kong ‘name and shame’ list.”
Are hedge funds bad?
I don’t know how they set the speaker order for the Ira Sohn Investment Conference, but it seems like a bit of a bummer to start with a guy whose fund was down 18 percent last year, 12 percent in the first quarter of this year, and who didn’t even have any new stock picks:
Larry Robbins, founder of the $8.2 billion hedge fund firm Glenview Capital Management, kicked off the Sohn Investment Conference in New York by imploring the 3,000 in attendance to ignore market pressures and see their investments through.
“Just hang on,” Robbins said Wednesday. “We just want to get from Point A to Point B, and frankly, as long as the fundamentals are tracking what we want them to be doing, we will get safely through our journey.”
Kicking hedge funds when they’re down seems to be thematic these days; here’s another article on the subject — “Hedge Fund Managers Lose Their Swagger” — in Bloomberg Businessweek:
Hedge funds have gone through tough patches before, but this one is disquieting, as the broader investing world isn’t in crisis mode. The S&P 500-stock index, after a rocky start to 2016, advanced about 1.3 percent in the first quarter. Hedge funds lost an average of 0.6 percent.
I suppose you could argue that the piling-on is pro-cyclical: Hedge funds underperform in a bull market, and will look worst just before the bull market runs out of steam. So all the hedge fund criticism might be bullish for hedge funds and bearish for stocks. Or, alternatively, disappointing hedge fund performance might be because crowding has competed alpha away, and things won’t be getting better.
Imagine if you had unlimited money. Not to spend — though that would be nice — but to invest. A lot of the big problems of investing — surviving drawdowns, calling bottoms, knowing when to cut your losses, keeping cash on the sidelines to grab opportunities — would just go away. You’d have some key advantages in terms of patience and certainty. “The market can remain irrational longer than you can remain solvent,” people would say, and you would laugh at them. “Not me,” you’d reply; “I can remain solvent forever; it’s kind of my thing.”
There is some loose sense in which this is the logic of the crisis-era bailouts, at least in the U.S., where Treasury and the Federal Reserve teamed up to use their essentially unlimited money and patience to buy financial assets at fire-sale prices when no one else was able to buy — and got themselves some pretty sweet (and litigated to this day) deals.
Another, smaller arbitrage that some people with limited money once tried is the on-the-run/off-the-run Treasury arbitrage: The most recently issued 10-year, or whatever, Treasury, tends to trade tighter than the next-most-recently issued 10-year, because people like to trade the most recent “on-the-run” issue and there is a liquidity premium. Eventually, though, their prices will converge, since today’s on-the-run issue will be replaced by a new one. So if you sell the on-the-run and buy the off-the-run, you should make a little bit of money, but to make a lot of money you need to lever the trade a lot, and if you are very levered you are very vulnerable to bad things happening. Long-Term Capital Management did this trade and bad things happened, though they were mostly caused by other stuff.
But the U.S. Treasury can do this trade with confidence. In particular, it can just make new on-the-run Treasuries, which it can sell at a premium, and use the proceeds to buy back old off-the-run Treasuries at a discount. It doesn’t even need leverage — or rather, for Treasury, selling the new Treasuries is how it gets leverage. It is a free-money-generating machine. The plan has other attractions. For instance: People are worried about bond market liquidity (see infra), and providing liquidity for off-the-run Treasuries would be a mitvah.
We talked yesterday about a proposal, mentioned in a Prudential comment letter to the Treasury, for Treasury to do just this; Prudential’s argument focuses on bond market liquidity and dealer balance sheet as well as on Treasury’s ability to decrease its borrowing costs. But apparently Prudential is not alone in thinking it’s a good idea; Treasury has been discussing it internally, though “the overhaul, which would be designed to improve trading conditions, has only been quietly discussed and could go nowhere as current Treasury officials are likely to leave office after November’s elections.” I don’t know. It’s such a Trumpian trade, honestly: It’s basically a debt restructuring where the debtor makes a profit.
Elsewhere in sovereign bond trades where the debtor makes a profit: “Almost $10tn of negative yielding government bonds are costing investors about $24bn annually, according to calculations by Fitch, posing challenges to long-term investors that rely on sovereign debt as a bedrock of their portfolios.” And elsewhere in suggestions for the government from the bond market, the Risk Management Association and the Securities Industry and Financial Markets Association think that the New York Fed’s Overnight Bank Funding Rate should replace the Fed Funds Open rate as a benchmark; here is the RMA/SIFMA letter.
People are worried about non-GAAP accounting.
Sometimes companies disclose financial results in compliance with U.S. generally accepted accounting principles, and also mention other, non-GAAP financial measures in the same press release. A lot of people dislike this. Gretchen Morgenson of the New York Times is a leading critic, calling the non-GAAP measures “fantasy math” and “phony-baloney financial reports.” I am less worried, since I am “both an efficient-markets fundamentalist and an accounting post-modernist.” Anyway, here is a MarketWatch story about how the New York Times itself reports “adjusted diluted earnings per share,” a non-GAAP figure that makes it look profitable, alongside regular GAAP EPS, which does not. Ha!
Morgenson said she can not comment on the paper’s use of non-GAAP metrics. A New York Times spokeswoman told MarketWatch, “Our view is that by reporting non-GAAP measures, we’re able to give our investors the best insight into the true nature of our continuing operations. These are measures that our management team reviews on a regular basis to evaluate and manage the performance of the Company’s business.”
People are worried about unicorns.
Yesterday I asked for unicorns playing ping-pong, and you answered. From @mileofyarn on Twitter:
“The corporate sector today is stuck in a vicious cycle of earnings management, questionable allocation of capital, low productivity, declining margins and growing indebtedness,” Druckenmiller said. “You can only live on your seed corn so long.”
Druckenmiller shared a graphic emphasizing the point. In previous business cycles, corporations have spent roughly evenly on what Druckenmiller called capital spending — research and development plus investments in office and plant equipment — and on financial maneuvers, including stock buybacks and mergers and acquisitions (M&A).
But since the financial crisis, Druckenmiller said, capital spending has increased $250 billion while buybacks and M&A have grown three times faster, by $750 billion.
The buyback-versus-investment debate is often framed as one between activist investors (who like buybacks) and corporate managers (who like capital investment), both because that is what is predicted by corporate finance theory (managerial empire-building versus strict shareholder monitoring of cash flows), and also because it seems to be empirically true a lot of the time. But it’s not inevitable. You could imagine corporate managers who love buybacks — because they are lazy or unimaginative or afraid of Carl Icahn or obsessed with short-term earnings targets that set their compensation or whatever — and investors, like Druckenmiller, who push them to invest in their companies for the long term instead.
People are worried about bond market liquidity.
Here’s Bloomberg Gadfly’s Lisa Abramowicz on a new paper finding that corporate bonds “trade about 3,000 times more slowly than U.S. government debt,” so it’s weird for corporate bond mutual funds to give investors daily liquidity. Maybe you should get daily liquidity but with T+3,000 settlement? Actually now that I think about it a 12-year notice period for mutual-fund withdrawals would not be the worst retirement-planning system, but we have wandered a bit far afield. No, Abramowicz’s view is that maybe corporate bond funds should charge investors a fee to exit their funds quickly, to make up for the trading costs that their exits impose on other holders.
If you’re new to the Raspberry Pi, that tiny $35 computer we all love, there’s a good chance you’ve downloaded (or purchased an SD card that includes) NOOBS. NOOBS makes getting started with Pi easy, and includes a bunch of different operating systems to choose from. Which you should choose depends, of course, on your project. Here’s how to make the right choice.
The so-called God Mode hack for Windows is rather less grand than it might first sound. Rather than granting users deity-like abilities, it simply provides one-folder access to an absolute butt-load of Control Panel options and settings. But security researchers have discovered that the technique used to create this special folder can also be exploited by malware. McAfee says that while the Easter Egg is great for power users, it is also being used by attackers for “evil ends”. By placing files within the God Mode shortcut folder, malware such as Dynamer is able to run undetected on a victim’s… [Continue Reading]
Original URL: http://feeds.betanews.com/~r/bn/~3/HKtHPP-eY_E/