Traditional enterprise IT operations carve up the management of IT systems into server, storage, network and database support teams. The suppliers of equipment for each of these domains are expected to produce the tools to manage the appropriate sphere of influence. “Über” managers then attempt to collect the data from each of specialized areas and make sense of the whole, usually with high implementation costs and limited financial savings.
In the earlier days of defining the business benefits of cloud computing, the emphasis was on equipment cost savings, both from better utilization of equipment by the service provider, and the potential to convert CAPEX into elastic OPEX. However, as the cloud model matures, it is becoming much clearer that cloud environments have to provide better productivity to developers and deployers of software and decrease costs of operational management. The only way to improve both the availability and management cost of infrastructure is to automate management processes. The old models of system management are changing quickly and radically, with the service providers leading the way.
Taking storage as an example, the way that service providers are providing management of storage is though APIs. For example Google provides a Google Cloud Storage REST-based API, as does Microsoft’s Azure Service. These sets of APIs can themselves be integrated to provide a stable basis for cloud management suites, such as OpenStack, VMware vCloud Director, Cloud Data Management Interface (CDMI) from SNIA, or even bespoke systems.
For the storage providers to play effectively in this space, they also have to provide the data in a way consumable by the ecosystem. The earliest example of this was the Cortex API system from Xiotech. This approach is also being followed by SolidFire, a new flash-only SAN storage vendor focusing on service providers. One particular requirement for service providers is to meet the service level agreement (SLA) for a client exactly. Providing too much resource sets performance level expectations that will lead to user dissatisfaction if they ever decline, even though the SLA has been met. Equally, over-providing resources leads to loss of potential revenue for additional service not being billed.
Xiotech and SolidFire provide detailed information and the ability to control the quality of service (QoS) metrics (bandwidth, latency and IOPS) to allow more efficient use of the storage resources by the service provider. This allows service providers to provide more tailored services for organizations at a lower cost.
In line with the Google and Microsoft models, both Xiotech and SolidFire are providing a management technology built upon REST-based APIs. Xiotech has included potential CDMI integration, and SolidFire have designed for deployment in a multi-tenant environment. This API approach allows for the automation of all tasks associated with deployment, provisioning, monitoring, and reporting within a storage environment. In turn, it is much easier to integrate storage management into cloud management suites.
Action Item: One of the key challenges of exploiting the potential of internal or external clouds is 100% automation of IT operational processes. The components within Infrastructure 2.0 need to play nice with each other and provide the data required to cloud management suites to provide full automation. That mandates that after a component is installed, there is no need to "log into it" for as long as it is deployed within the infrastructure network. The acid-test of fitting into cloud-scale management is full automation without exception.
Footnotes: