A lot of the pundits pontificating about the cloudpocalypse are blaming application developers for not building redundancy into their apps. I'm sure in some cases, that is true, but the reason this became such a disaster was precisely because app developers built in redundancy. The cloud failed because everyone expected the cloud to fail.
Amazon's EBS technology is fundamentally flawed. This has become increasingly apparent in the last year, maybe longer. As a result, developers have built in all kinds of mechanisms to recover from EBS slowness or failures. So as EBS fails, the recoveries kick in, which spreads the problem to other data centers, which causes those to fail, which then causes more recoveries, etc.
It's like the stock market and program trading. When the market starts to tank, computers automatically respond, pushing it even faster. To combat this, the markets have put in limits to slow down program trading. No such thing exists in the cloud, I don't know if it would even be possible.
But there's another parallel to the stock market. Whether it goes up or down is based on whether people think it will go up or down. We're now in the situation where even more people will be building increasingly sophisticated recovery mechanisms for cloud failures, which will only make the problem worse.
If the stock market is too volatile, the only option is to put your money some place else. I've decided that Amazon's cloud is now too volatile, and I have to move my assets elsewhere.
Showing 6 reactions
What are you thinking for a hosting solution? Dedicated? I’d love to chat more about strategies for migrating out of the cloud, I’ll shoot you an email.
Cheers,
Aaron



