3 ms·
The difficulty of scaling is not in getting the hardware, but in the operational burden and frequency of failure. With EBS, machine and disk failures are effect
by tybris 15y ago
The difficulty of scaling is not in getting the hardware, but in the operational burden and frequency of failure. With EBS, machine and disk failures are effectively irrelevant and you can instantly migrate to another machine. You don't need to worry much about data loss. If a disk breaks in one of your machines, EBS keeps working. If the machine breaks, the data is still in EBS. If EBS breaks, the back-up is still in S3.
Dedicated hosting and rackspace cloud may be cheaper, but if you expect to be in a situation where failure becomes frequent, you'd better be on EC2.
- nirvana 15y agoWith AWS, the development, operational and financial burdens are higher, and the frequency of failure is much higher. EBS is not a unique feature, as there are multiple equivalents of EBS, along with higher level methods of replicating data, namely, couchDB, riak, cassandra, etc. But I think you've answered my question- people use AWS because they don't realize this.
- tybris 15y agoIf you are starting up, where are you going to get the capital, expertise, and people to operate and maintain those systems? You have better things to do.
- nirvana 15y agoI love it when people tell me spending more money and time and effort on something that is less flexible, reliable or performant is somehow saving money, time and effort. I think you do not understand what you are talking about and you're giving me marketing spin, in fact that must be the case. You're simply wrong. I've explained it, you're not listening. Of course the kicker is that this is in a thread about how a bug in Amazon means that the EBS volumes that were allegedly being backed up, actually weren't.