The development of many large-scale Web 2.0 applications is being built around standard hardware and software components, driven by the imperative of reducing the cost of development and operation. The hardware components are Intel architecture servers, IP switches, and SATA drives. The software components include open-source Linux OS and files systems. The applications are very lean, only using the software components necessary to enable the application. Written code is kept to a minimum.
The architecture of an IASS environment assumes that software and hardware components will fail and fail often. Redundancy and recoverability are built into the architecture (often in the files system). Hardware components are hot-swapped. There is no hardware maintenance.
What's missing from this scenario are the traditional storage arrays, storage software, SANs, VMware, and application middleware such as Oracle. Software licenses on a per terabyte or per CPU basis just don’t cut it in this environment.
Clearly IASS is not suitable now for high-performance transaction environments. However, for many applications with large-scale tier-3 storage they represent budget reductions of 50-80%.
Action Item: As storage systems such as EMC’s Hulk come to market, IT executives should expect to see IASS from MSPs in the data center for applications such as archiving. Senior executives should be challenging their development teams to come up with alternative architectures for tier two and three applications, and driving up the adoption learning curve for IASS.
Footnotes:
