Slashdot Mirror
Amazon and Hardware As a Service
sioux_chance writes to recommend an article up on ReadWriteWeb comparing Amazon's S3 and EC2 services with Google AdSense. (They are not the first to coin the term "HaaS" for hardware as a service.) The analogy is that Google increased the granularity of (the article invents the term "fragmentized") the revenue side of the Web business, whereas Amazon's HaaS does the same for the cost side. A comment to the blog posting points out that NearlyFreeSpeech.net has been selling fine-grained hardware capacity for years, but Amazon does bring a greater scale to the business.
For many decades, IBM only rented machines; they didn't sell them. Not until they lost an antitrust case did they sell hardware. Rented machines came with IBM service, which was excellent. Now that was "hardware as a service".
What Amazon is offering is called "time-sharing".
Remember Sun's "grid computing"? Big dud. The number of people who want to pay to run huge batch jobs but don't want to buy their own hardware just isn't that big.
There are two players in this space who are known to make money: Akamai and ResPower Render Farm.
the hardware isn't yours, the software isn't yours and if the censoring is any indication your data isn't yours either.
http://www.thefreedictionary.com/fragmentized
fragmentize (frgmn-tz)
tr. & intr.v. fragmentized, fragmentizing, fragmentizes
To break or become broken into fragments.
The American Heritage® Dictionary of the English Language, Fourth Edition copyright ©2000
s3 is primarily geared towards file storage, so probably a good choice for hosting static files, images and CSSs. but what about relational data - hosting real databases. There is no persistent storage in ec2, making any virtual server setup extremely unreliable.
That said, s3 makes sense for dumping things like tier-3 data backups.
95% of all sigs are made up.
Akamai is a different beast altogether -- they're not a generic computing center, but a finely-tuned cache that gets your (mostly static) data as close to the user as possible. In fact, Amazon uses Akamai (PDF press release) to host some of their content.
The difference between Sun's Grid Computing and EC2 is that EC2 is connected to the net. This doesn't mean you can't run huge batch jobs on EC2; however, there's a lot more you can put on there (read: hosting for the Web 2.0 company you've founded in your garage, mom's basement, ...). However, EC2 doesn't give you a load balancer (yet); getting the traffic from www.your-spiffy-domain.com to the EC2 instances is still your problem.
S3 is, IMHO, the more interesting of the technologies today. Buying storage capacity these days is cheap; maintaining it, however, is as expensive as ever (perhaps moreso as clients expect higher availability, geographic distribution to minimize risk, etc.). And, if I'm too small for Akamai yet need to host some static content over a fatter pipe than I have, I can even expose it to the rest of the world through the REST interface./p.
you really couldn't find a better way to contribute to the discussion?
your better way is to bitch about his lack of a better way?
I've heard this before and didn't understand it.
What does "no persistent storage in ec2" mean?
There are two players in this space who are known to make money: Akamai and ResPower Render Farm.
Oh, come now. That's just silly.
When you host a website, you pay for NNN Megabytes of website, and maybe YYY GB of network transfer. These are terms of hardware, not software, not service. Yet, it's sold as a service.
Even the free Hotmail has, as one if it's key features, XX GB of space. It's a free service sold in terms of hardware consumption. Then there are backup providers, weblication providers, Yahoo stores, etc.
There aren't two, there are something like 5 million of them.
I have no problem with your religion until you decide it's reason to deprive others of the truth.
Amazon S3 & EC2 are revolutionary, but at some point, it's a reasonable next step. The only big drawback of EC2 is the lack of persistence so it's hard to host a dataserver on there.
But the truly revolutionary service is Mturk. It's about packetizing tasks for humans! not for computers.
Your Sun's "grid computing" link has nothing to with Sun or grid computing.
I looked at S3 with an eye towards using it as a backend for a hosted backup system (one of their recommended uses, in fact). The problem that I had with using it is that there is no transparency about their operations, no service level guarantees and really no information about just where things will work and where things will start to break down.
For example, the talk about S3's scalability as being "unlimited". In reality, nothing is unlimited. They might have an architecture that theoretically scales up indefinitely but the implementation could be quite lacking. I could see two ways to use S3 as backup storage - you could make a tar file or equivalent and put it into S3 or you could assign each individual file an S3 ID and put it in. The difference between the two approaches is huge in terms of the number of unique objects stored (I have over 1.5M files on my laptop. Now multiply that out by 10's of 1000's of users). Designing a system that can stores 100's of billions of unique objects is different from designing a system that can store millions of unique objects. Not knowing what S3's intended workload is, it's hard to design for using it.
Furthermore there are no guarantees about bandwidth, availability. There aren't even guidelines for how much they think is reasonable. If I were to design a service using them as a backend and then have big success, what happens? Even if they're willing to add capacity at a rapid rate, how rapidly can they add? If a customer calls and says "I can't access my data" - how does that get resolved? How global is S3? If I start picking up customers in Europe or Japan or China, what will happen?
So, I think S3 is a great service for small uses. It's not possible to build a larger business around it until they start being more transparent about how it works, what are reasonable uses for it and how it is being used currently.
As a developer, customers have asked me why I don't add S3 as a backend to my backup app. That would be reasonable, but what's in it for me? S3 doesn't have any kind of referral fee so I'm just handing them money and adding headaches for myself. If they want people to add it their apps they should set up a way that the developers can get a little piece of the action.
I hope they set up some kind of system preventing the data used in the service from falling into the wrong hands, and by that I mean something above par. It will only take one successful hack of the system to make their business model come crashing down. Most of their potential business clients will only be willing to pay if they can assume that their data will remain secure with them as the only ones who can use it. Don't get me wrong I love the idea(although it is nothing new) but I hope they think this through.Just like windows is a prime target because of the potential loot this will likely be targeted.
I can see it now:
"Breaking News Hackers have Secured Top Secret Files Proving [Conspiracy Theory]! [Government Organization] Denys All Claims"
"Amazon Bankrupt After Lawsuit when Hackers Steal Client's Valuable Confidential Files"
"Slashdot Users' wetdream- Source Code for next Windows Operating System stolen off of Amazon's HaaS after Microsoft Budget Cuts due to the Well Known Vista Blunder"
The word is SPLIT, it's much clearer and aesthetically pleasing then "Fragmentized"
We looked at using Amazons storage thing for our start up. There are no technical reasons we didn't use it, we decided against purely because Amazon have been such assholes with patents. Why add to their profits if there's a chance you'll be sued for some obvious technique 5-10 years from now?
Just have to put in my two cents (which would buy you 2 MB/month storage or 21 MB of outbound traffic) in professing my love for NFS. I don't think I've had a bill over 50 since I went with them.
Welcome back to the future (c) 1970-1980. HaaS or SaaS is just a re-invention of time-sharing. Don't need to configure your computer, we will do it FOR you. Don't need to buy software, we will do it FOR you. Don't OWN your software/hardware, WE do. Now comes the slow push of Don't store your data, let us do it FOR you. Catch is, you no longer OWN anything, not even your data.
Need something that can still be re-used for a few years? buy a computer.
Want to throw away money to a corporation and make them richer, buy into this bullshit model.
The summary mentions NearlyFreeSpeech.net which we use to host paid research/Q&A site http://uclue.com/
What we like about them is that we pay for what we use. No more, no less. Why is this concept so rare in the industry, which seems to be built around "pay for promises, get what the arbitrary fair usage policy gives you"?
The downside of NearlyFreeSpeech.net is that they don't offer https due to some ideological problem with IP Addresses. The upside is that, apart from that, they make money by enabling you to do more, not by restricting you from doing so much.
Paid Q&A/Research
Google's Adsense makes something accessible to small players that is hard to do otherwise - soliciting and managing business relationships with advertisers and web sites, and integrating an apparently fair revenue model into it too. Amazon's HaaS isn't that radical/new. For one thing, fragmented computing/networking resources were already available as Virtual Private Server or Virtualized Environments. Perhaps not as fine-grained but cheaper than Amazon.
Given Amazon's linear pricing, their granularizing everything makes me nervous... how often do I have to watch my meter as I run my apps? A VPS with fixed limits on everything except total bandwidth used, at its price point, is something I'm more comfortable with at this point.
The biggest and mostly white one of them all, the US government
:P
and stop trolling
I'm getting an entire server (not virtual), with hardware RAID mirror and other reasonably decent low-end server hardware (2 GB RAM, Core 2 Duo). I can do whatever I want with it, and the provider supplies the hardware, the O/S and the bandwidth (4 MBytes / second, up and down) for $260 / month. If the hardware breaks, the provider fixes it within 4 hrs. For extra money, I can get more fetuares, like mirrored servers, etc... For an extra $300 / month, I can have 10 M Bytes / second symmetric. They set it up for me within 48 hrs of order. It's been great so far.
of business puut and end to thaat name?
Previously: "Linux... Toward the Sunrise..." Now: "Linux... Toward the-- No, now, part of Every Sunrise"
Your Sun's "grid computing" link has nothing to with Sun or grid computing.
Oops, sorry. Pasted in a link from something I was doing in another window.
and is your way ... oh fuck it! Let's get BACK on topic, shall we?
I, for one, get on topic and welcome our word-reinventing overlords.
Now, can we discuss the article? This goes for gramer nasies 2! LOL!
Amazon has other services to help keep state with EC2
SDB -> Simple Data Base
SQS -> Simple Queue Service