Am I wrong in wondering if cloud computing might do for computing what hedge funds and CDSs did for finances? By this I mean that in both cases we're talking about attaining the last iota of efficiency by consolidating a resource in the hands of a few. What happens when the MS cloud goes down? (I know. Impossible.) We've seen the stories about people who have lost access to their gmail accounts. What if a large portion of the economy has their applications tied up in the cloud and there's an interruption in service. Or a natural disaster.
Analogously, should these cloud providers be required by law to maintain a (geographically distributed) computing reserve? A second (or third, even) clone of the current cloud?
I understand that the communication between the cloud and the world is distributed, but the actual cloud seems to be as distributed as a rock.
Amazon currently have 3 separate data centers and additional S3 storage in Europe. This trend will continue as SLAs become more of a selling point for all Cloud purveyors.
Second why do people think that your own data center will be better? It is like putting your money under the bed rather than in the bank. So people do that but not a good idea.
Lastly we will see a common API emerge that will allow us to move with little impertinent. The emergence of Eucalyptus is an early indication of this.
Your reply seems to be a spot-on agreement with my post, except that you don't realize it. Your choice of analogy is particularly amusing in that, right now, if you'd had your money under your mattress, you'd be better off than if you'd invested in one of these highly-leveraged and over-exposed financial vehicles that I'd mentioned. Mattressing one's money may have been the best idea of them all for a large class of investors.
That aside, the proper way to make the analogy you're making is not to compare investing in a CDS against putting money in or under your mattress, but rather to compare it with putting it in a certificate of deposit in your locally-owned bank or loaning it to a locally-owned business. Either of these would very likely have less return and a higher risk for you. However, you would be less susceptible to a global default cascade like the one that is happening now.
The reason that ...your own data center [would] be better... isn't that it's less likely to go down, but rather that, if everybody has their own data center, the chances of a global blackout go down drastically. This is precisely the analogue I was pointing out. If banks had stayed non-optimum and local, then (arguably) we would not be facing a (potential) global economic meltdown.
You may argue (rightly) that if you pursue these localist strategies when everyone else is pursuing these more profitable consolidated strategies, which carry this systemic risk of catastrophic failure, that you'd succumb to competition from others. This is the point of asking for regulation in the market, and (possibly) the case for asking for it here. At some point, almost by definition, the cloud computing market will commoditize. When that happens, the only way to cut costs will be do do without redundancy, if that is a legally available option.
The underlying problem is that people discount the cost/benefit of events that happen in the distant future, and the global catastrophes in these two systems always seems distant.
Finally, one may object that if redundancy is such a good idea, then it will happen naturally and never be eroded by market forces. In this case, then, there can really be no objection to legislation on the matter, because everyone will naturally comply anyways.
Almost nobody has yet lost money on typical retail bank products. Even a checking account is better than the mattress. Retail banking is doing just fine. Not that that's too relevant.
They did... Until the Icelandic government reneged. The UK government (lots of UK citizens had money there) then seized all their assets in the UK. It's a huge diplomatic mess.
One question that comes to mind: How does this effect the market for private, dedicated (managed) servers? Or rather the platform providers for today's online apps?
The root website is www.azure.com - that should link to all the relevant docs, bits, sites. We're turning on all the switches as I write this so you may see a few 404s before noon PST.
I would be hesitant to start using this because there is no way to anticipate the pricing scheme whenever it shows up. Last thing I want is to commit to a platform andthen find out it's cost-prohibitve. Also, google promises to keep certain level of usage free which makes me more likely to try because I know I won't get suddenly hosed with high prices.
I've heard this come up many times today. I'm curious if everyone questioned the financial motives of Google App Engine when it first came out.
Then again the target market for Azure seems to be enterprise IT depts, who tend to be involved with proprietary technology, and not start-up developers looking for a cheap and efficient host.
1. Yes, I did question it back then too. Google has since rectified their stance and published their pricing scheme. The paid service is not operational yet but it's a neccesary piece of inforation to plan the projects ahead. I have started learning pyhon based on that information (python is required for google's services) but did not commit the project itself yet because the ship date is still up in the air.
When did you actually trust Microsoft with online computing? Before Vista? I don't see a huge relationship between Vista and their online efforts. Sounds like a bit of MS bashing propaganda from somebody who has never liked or trusted them.
Not entirely true. They've really gotten it where console gaming is concerned, moreso than anyone else. The Xbox has been consistently better than its rivals at that.
They're now serving up millions of dollars worth of games per month (allegedly 70% of Xbox owners have used Live Arcade) and are moving into streaming video too. I can now watch MSNBC through it.
They've achieved the dream an internet-connected box on millions of people's TVs. Apple tried and failed miserably with their TV product.
There clearly are people at the company that get it.
The Xbox is not the internet. It is a dedicated internet-powered appliance. And while they may have done a good job there, they have failed in every other internet initiative.
Microsoft does not understand the internet and has never made any notable internet service. Silverlight? Live? MSN? Hotmail? All knockoffs. And they have "failed miserably" every time.
Why would I trust their new internet cloud system when it directly, extricably integrates into all of these other failed initiatives as a feature?
I don't get what you mean. They've failed at search, clearly, but so has everyone but Google. Hotmail is most definitely not a failure, and was a pretty smart acquisition. Silverlight is too new to call a failure, though it's clearly not going to be taking much market share from Flash any time soon. If by MSN you mean their instant messenger, it's the largest by far.
Some huge sites (Myspace, MS's stuff, PlentyofFish) run on their servers and programming languages. Their browser still has the majority of the market share (though arguably not for quality reasons.)
I don't understand this "they're failing at the internet" sentiment. It seems to just be normal anti-Microsoft stuff. They're failing at search, not the internet as a whole.
Everything they do is a knockoff of something else, and it's usually not better (I actually cannot think of a single case where it is better). I am not just randomly anti-Microsoft for religious reasons. They make terrible products that have horrendous user experiences. They do not innovate. In general, I have heard they hire sub-par programmers.
All of this is exaggerated in their internet initiatives. Who uses Hotmail anymore? Who used MSN, which they abandoned for Live? Who uses Live? Why would they try to compete against Flash with Silverlight? What gives them the authority or experience to do such a thing?
And if you asked me which were the two most horrendous monstrosities of popular websites on the internet, I would say PlentyOfFish and MySpace. They are the two sites which have the least sense of design, ergonomics, usability, and general respect.
Why, then, should I trust that they would make good decisions on their backend code and server software?
adcenter is their answer to adwords/yahoo search marketing(overture).
I'm still in shock that the xbox doesn't have a browser. That's really scary. Either they are just amazingly arrogant and want everything locked down walled garden, or they are still in denial about the importance of the web, or they just don't get it.
Idk. Wii has a web browser and it sucks and nobody uses it. I purchased it out of curiosity, and what a waste.
Without a reasonable keyboard, there's not point. And reading the internet off the TV sucks. I'd rather surf on my Moto Q, and that's saying a lot because surfing on that tiny thing sucks ass.
I'm not sure how you're measuring console success.
For me, Nintendo is the clear winner of the current console wars. Nintendo makes hideous amounts of profit on everything. XBox as far as I understand still makes a gargantuan loss. It's a long term strategy, and who knows, maybe it'll play out for microsoft, but I don't think it's relevant to the question of ms understanding the internet. It just shows they understand getting people to use something and taking a massive loss on it.
The Xbox division is now actually turning a profit. Not enough to offset the massive losses it's had in the past, but it is actually generating revenue now.
The Xbox itself is a good product. They overlooked the market the Wii pursued, but then again everyone but Nintendo did. If I recall Nintendo has never had an unprofitable quarter, even during the GameCube days when they were doing much less volume than the competitors.
They've always viewed it as an investment. Honestly, it's worked out pretty damn well. I'd never have guessed they'd have gotten to 2nd place so fast.
Moreover, they've just begun to capitalize on it. As I said, they've got an internet connected box tethered to everyone's TV. They've been working on the dream of the living room PC for years, and now they've finally gotten there.
It's a huge deficiency. On the other hand, the voice chat, friends lists, instant messenging, game matchmaking, and online marketplace all work pretty well. Nintendo has a browser and an effective marketplace, but the actual game connectivity is lackluster.
The DS has been astoundingly successful also. Especially in getting non-gamers to buy it and use it.
Nintendo played a blinding strategy - let Sony and Microsoft play "Who has the biggest willy" with their hardcore gaming hardware which renders 4 gazillion polygons a second, while Nintendo makes fun games with wide appeal.
How about discussing the proposition on its merits, rather than pedigree?
It's not like there are ten new cloud platform come in every day and we have to make a snap decision. We can, in fact, afford to look at the offer itself.
Sorry, but if I want to buy a laptop, I wouldn't buy one from Pepsi. Similarly, if I want something Internet related, I wouldn't get it from Microsoft. I just don't see any evidence they understand it. They have done an absolutely awful job in the last 10+ years.
If I want to buy something development and execution architecture based, I'm happy to buy that from MSFT since they've done exceedingly well there. How many sites out there have been running on google and amzn's dev/exec arch's the past 10 years? A lot less than Microsoft's. As far as them getting the internet or not, who cares. There's more to life than the internet. Java and .NET run the enterprise, not sure who among MSFT, IBM and Sun really "gets the internet" in the sense you're probably working to define.
I'm interested to see how well the service fairs when it launches and I'm just as interested in seeing the response from those developers who are being able to test the platform. I think that this could potentially be a "step in the right direction" for Microsoft to stay relevant in a world that's changing how and where data and processing is handled. However, unlike many speculators who talk about how someday the "Cloud will completely engulf the world," I don't think that will happen anytime soon. As available as the Internet is and the number of providers grow, there still is going to be a need for localized data processing and handling to a certain extent. This is one of the places that I agree with Microsoft in saying that it needs to be "Local + Cloud." The concept of it all strikes me as something best used to enhance local applications to have access to more information, seamlessly, and dynamically, and be able to tailor that information to meet the needs of the end-user.
Definitely interesting! From what I see at first glance, there's a queue/MQ-like service, a scalable database (is it SQL or key/value?), a workflow service, and the whole thing works whether you want to use .NET or not.
You can find more docs on Windows Azure tables (distinct from the blobs service which you missed :-) ) at http://go.microsoft.com/fwlink/?LinkId=131257 . Sorry about the doc link. I'm sure there's a web version somewhere which I can't find right now thanks to the flaky connection I'm on
Maybe I'm missing something here, because nobody is mentioning it, but isn't this a major shift for MS? Microsoft is a software company basically (other than mice and zunes and 360s), but here they are offering essentially hardware-as-a-service. If, last week, I would have gone to Dell and bought a server with a windows license (Disclaimer: I'd never do such a thing!), I'd now just pay MS to run my app on Azure . Dell, HP, IBM and whoever else make windows servers have just been cut out of the equation, right?
H/w vendors are only cut out if MSFT invents a replacement for metal. Granted, utility computing in the cloud is about squeezing more utilization and value out of a given piece of h/w, so it's not without impact. You might perceive this as a change, but I'd say it's a change in the paradigm moreso than a change in MSFT.
Also, people tend to overlook this but cloud isn't really about infrastructure or hosting. Sure those are the current analogs--cloud is about what happens next, and I think it's more about shifting architectural tenets so that we do what works well with cloud scale and lifecycle. When you do those things, other interesting things start to become possible (e.g., in the cloud, it's hard to have misaligned test and prod envs, not the rule like it is most everywhere else).
Analogously, should these cloud providers be required by law to maintain a (geographically distributed) computing reserve? A second (or third, even) clone of the current cloud?
I understand that the communication between the cloud and the world is distributed, but the actual cloud seems to be as distributed as a rock.