We think that people are choosing to use the cloud because theyve found building their own reliable infrastructure difficult, expensive, and in the end, well, unreliable, Kinton said.
Computers, networking equipment, and even whole datacenters can and do fail. All computing systems have to deal with failure. . . In most cases, given the focus we put on this and the scale at which we can operate, the cloud can provide better tools for insulating customers from failure than could reasonably be engineered by any single customer. In essence, one should design for failure as a normal and expected part of operation instead of as a rare and surprising event.
For example, an Amazon EC2 instance has the same failure characteristics as does a server in a datacenter. When an EC2 instance fails, replacements can be instantly provisioned through an API call. Alternatively, they can run concurrently for redundancy. Amazon EC2 also provides Elastic IP addresses. These allow customers to remap an Internet address from a failed instance to a working one.
Because tornadoes, fires, or a variety of other disasters can, and do, hit even the most hardened data centers (or their connectivity), EC2 instances can be launched in different Availability Zones separate pools of infrastructure with independent location, power, cooling, and connectivity, Kinton said.
Amazon S3 stores objects in multiple Availability Zones, yielding durability beyond that of traditional RAID storage with an onsite tape backup. Amazon SimpleDB Domains are stored the same way. Weve also recently released new features that provide monitoring, auto-scaling and elastic load balancing for even greater resilience in building applications.
Whether you believe Amazon Web Services is ready for primetime or not, one thing is clear: theyll need to fend off plenty of competitors to stay on top.
First, there are the other Fortune 500 players, such as Microsoft, Google, Oracle and IBM crowding into the space. (IBM recently entered a cloud partnership with Amazon, which muddies the waters a bit.)
Then there are the major Internet players like Salesforce.com, Akamai and SAP. Finally, a slew of startups loom on the horizon.
The cloud space is so crowded (or the term is being used so promiscuously) that there are now scores arguably hundreds of cloud computing vendors. Plenty of vendors, start-ups and organizations claim to have a cloud presence.
However the space shakes out, Amazon intends to plow ahead with its cloud vision.
[An] interesting data point is that the bandwidth for just two Amazon Web Services (Amazon EC2 and Amazon S3) has exceeded the bandwidth for all of the Amazon.com global websites, Kinton pointed out. Additionally, our storage service, Amazon S3, now holds over 64 billion objects and regularly peaks at 100,000 requests per second.
That sounds like a pretty good head start on the competition to me. Amazon clearly intends to develop a cloud ecosystem, along the lines of what Salesforce provides around its main offerings, which is pretty much table stakes these days if you want to play in the big leagues.
Recently introduced services include one-on-one support options; beta testing of AWS Import/Export, which helps accelerate the movement of large amounts of data into and out of AWS using portable storage devices for transport; a pay-as-you-go content delivery service; and Amazon Marketplace Web Service, which allows sellers to exchange programmatic data with Amazon which will ultimately help sellers gain efficiency and grow their businesses.
Amazon may not quite have a cloud empire, but its more than ready for battle.