Slashdot Mirror
EC2 Outage Shows How Much the Net Relies On Amazon
An anonymous reader writes "Much has been written about the recent EC2/EBS outage, but Keir Thomas at PC World has a different take: it's shown how much cutting-edge Internet infrastructure relies on Amazon, and we should be grateful. Quoting: 'Amazon is a personification of the spirit of the Internet, which is one of true democracy, access to the means of distribution, and rapid evolution.'"
An article at O'Reilly comes to a similarly positive conclusion from a different angle.
This article seems to be an apology for Amazon.
Basicly it says "We went down, and took down lots of important stuff. That shows just how important we are and that lots of people use us. Thus, our cloud is a good thing."
The logic of that doesn't quite work.
I agree that it's a useful tool, but there are a lot of things that don't make sense to put in the cloud.
Many .com websites were unnecessarily down for hours since nobody had thought to plan for a outage. I am sure quite a few architecture meetings where held the following day addressing disaster recovery.
Got Code?
Wait....we should be glad we have a single point of failure on the internet because why?!?
"We can categorically state we have not released man-eating badgers into the area." - UK military spokesman, July 2007
I guess that the major difference to traditional outsourced hosting is what you mentioned but didn't emphasis... The "scalable" part. If you normally spend X amount of resources (CPU time, memory, whatever) and might get a peak of 50X resources at some point, traditionally you would either constantly pay for a lot of resources that you didn't need for most of the time, or your service would crash during the peak. Cloud offers a lot more flexibility as you can pay based on what you use, not based on what you estimate you might need. Pretty useful for some things, though certainly overhyped (and because of the hype, some have reacted with the "It's useless!" attitude, which is just as wrong).
Disadvantages are pretty obvious: Your data is at the hands of a third party.
The cloud represents a black box that abstracts the underlying network topology.
You might send your data to a server in Germany and retrieve it from a server in the USA. When you put something in the cloud you do not have to worry about problems like this because the cloud provider already has a hot backup ready to take the slack in another part of the world. You don't need to know or care how it happens, it just works. S3 is an Amazon example of a cloud service. You send your file to S3 and Amazon takes the responsibility of ensuring that it is available even if a datacenter is blown to smithereens.
EC2 and EBS are not the cloud. There is no abstraction of the datacenter. Amazon leaves it up to you to choose which datacenter you wish to work in. This can allow you to easily build a cloud application on top of their physical infrastructure, but it is up to you to make it "the cloud". We witnessed so many failures because the applications were not cloud applications, just standard hosted services.
Microsoft: We're sorry our product broke and a lot of people weren't able to get online. Slashdot: BURN THE HERETIC! Amazon: We're sorry our product broke and a lot of people weren't able to get online. Slashdot: It's okay. Here, have a cookie.
#fuckbeta #iamslashdot #dicemustdie
Paying for multiple availability zones is not the same as paying for multiple locations. There are multiple availability zones in a single datacenter. Netflix got it right, they spread their infrastructure over multiple physical locations, and didn't suffer any downtime despite losing a significant chunk of their infrastructure; it was business as usual.
Like anything else, cloud computing still requires you to decide how much redundancy you're willing to pay for. If uptime is that important to you, spreading your infrastructure out over multiple datacenters is a no-brainer.
"'Amazon is a personification of the spirit of the Internet, which is one of true democracy"
I'm sure Wikileaks would disagree.
Amazon is a personification of the spirit of the Internet, which is one of true democracy, access to the means of distribution, and rapid evolution
Spirit of the internet? Some on seeing Amazons' passing judgement on Wikileaks might think it more aligned with a certain corporate spirit than a spirit of the internet. If they're really support democracy, which can't function properly with a poorly informed public, maybe they shouldn't be the ones to decide whether or not someone is a journalist.
Hardware doesn't make spirit. What people are doing, and the thoughts that drive the choices made probably do.
They are still contented to profit from the sale of books about WikiLeaks.
http://www.amazon.com/Inside-WikiLeaks-Assange-Dangerous-Website/dp/030795191X
http://www.guardian.co.uk/technology/2010/dec/11/wikileaks-amazon-denial-democracy-lieberman
I was directly affected by this outage. Once i discovered that the issue was at amazon and not at application- i restored from a previous snapshot, synced my application code, and associated my IP to a new instance in a functioning zone.
Total downtime for me was probably just under an hour. And that's including my debugging time.
Overall it wasn't the end of the world for me and i did discover I should make my redundancy setup run more frequently.
Sure i lost a few sales, but in a way i look at this as an example of why I should be better prepared for such an occurrence.
This still isnt as bad a when IBM pulled the wrong drives out of my server and wiped them.
With EC2, they can scale automatically and programatically and can spread the virtual servers across multiple regions for additional redundancy. All with a single API.
That sure as fuck didn't seem to be the case these past few days.
All my websites are fine, which is what my high profile clients expect.
That's because we use Microsoft Windows Servers and Sql Databases.
Really? I've found both such products to be unsuitable for the demand we put on such infrastructures - unless I throw a lot more hardware at them. With 1/20th the traffic, and 6% the userbase, our forums crawled on Windows Server and MSSQL Server. We switched to Apache and MySQL, and even running the greatly more database intensive (than the Windows solution we were provided) Simple Machines Forum, we need a lot less hardware than we previously did when we had so much less traffic.
StarTrekPhase2 - The Five Year Mission Continues!
Don't forget the one-click patent. True democracy/spirit of the Internet my ass.