Today I was reminded that no matter how much we take technology for granted, performance and scalability are still a big challenge. Case in point: Put 9,000+ software, network and hardware
engineers into a Las Vegas conference facility and internet access gets scarce. There’s upside to it, however —that endless stream of work emails gets diverted.
Day 2’s keynote enumerated the reasons why people are and should be using AWS, as well as showcasing partners that take advantage of AWS’s many benefits. According to Amazon (Web Services) CTO, Werner Vogels, AWS delivers value and innovation in:
A few announcements that came on the heels of these “core values” for AWS included:
- Support of PortgreSQL on Amazon RDS
- Ability to move data between AWS Regions for increased system reliability
- Amazon Kinesis, a high performance, fully managed service for processing stream data in real time to support the growing needs of various data processing applications, such as processing
tweet streams, sensor data, etc.
- G3, HI2, I2, and C3 instance types
The list of partners was pretty impressive, as well as what they do with the technology. Most of the following companies wouldn’t even exist if not for AWS’s cloud infrastructure and
- Cycle Computing
AWS has allowed them to scale to extremely impressive numbers. For example, Cycle Computing ran a computation simulation involving 16,788 EC2 instances in eight regions and completing 264
months of computing in less than eighteen hours. Building out that kind of capacity would have cost them $68M, but they did it for $33K on AWS! And Amazon demonstrated once again that it is not afraid to showcase partners that offer competing services, such as Parse and its MaaS offering.
The highlight of the day for me was when the Amazon.com Data Warehousing team discussed their experience migrating portions of their existing Oracle RAC-based Enterprise Data Warehouse to Redshift. (Their EDW is one of the largest data warehouses in the
world, in multiple petabytes.) While they were pretty clear that they are not planning – and it is probably not feasible – to migrate completely to Redshift, their experience demonstrated that
they can move about 45% of their EDW there. Plus, the workload moved thus far shows a cost reduction of about 55% on most jobs.