The most impressive thing about VMworld 2009 was the clarity of vision, both of what business problems VMware is trying to solve and the future architecture of VMware to solve these problems. Currently VMware can solve three problems well:
- Poor server utilization - VMware reduces CAPEX for the server infrastructure;
- Poor availability for Tier 2/3 applications - VMware allows a much simpler restart paradigm than alternative solutions such as Windows clustering;
- Lack of rapidity and flexibility of provisioning new servers.
Paul Maritz has focused VMware on providing the future "Software Mainframe" infrastructure based on virtualizing and encapsulating application environments on commodity hardware and software. The key business requirements addressed by this aggressive vision are the ability to:
- Move applications/volumes/blocks dynamically within and across arrays to meet performance SLAs and minimize cost;
- Move applications/volumes to meet availability and DR objectives (fail-over & fail-back) within & across arrays at distance;
- Provision storage rapidly, dynamically and automatically to reduce administrative cost and increase availability;
- Migrate storage dynamically within and across arrays, saving CAPEX, increasing financial flexibility and reducing administrative cost;
- Secure, archive & delete data to be compliant and secure;
- Report on costs and SLAs by business group and by application group in order to bring costs and accountability closer to the business;
- Work as effectively in virtualized as in non-virtualized environments.
The road to achieving this vision is not easy, and certainly much longer than the two-to-three years suggested by VMware executives. This virtualization and encapsulation strategy produces complexity and overhead in itself, even as it attempts to reduce complexity overall. For example, performance problem determination requires substantial expertise to diagnose and remediate.
Other approaches to meeting these requirements are being put forward by vendors. For example, Oracle (the Sun integration delay not withstanding) and Microsoft are creating stacks based on an application view of the world and embedding the functionality above under the control of the application.
Information management and storage vendors, among the important players in this space, are under significant pressure to improve their vision. If they want to participate in the enterprise infrastructure, they need to be able to offer all the functionality above. Storage vendors have great fail-over DR products but limited fail-back. Migration to new arrays and de-commisioning of old arrays is still very expensive, with long project elapsed times.
Action Item: To maintain margins and relevance, IT vendors must provide more automated data movement, migration and security functions, either alone or in partnership. IT executives should demand clear road maps on how and when vendors plan to deliver these functions.