Data Center of the Future – now
photo by youflavio
In a recent blog post Eric D. Brown defined an Agile Data Center as
An Agile Data Center is one that allows organizations to efficiently and effectively add, remove and change services at the speed of the business, not the speed of technology – Eric D. Brown
In follow up post he said that a Agile Data Center could be implemented by Software Defined Data Center (SDDC) for example using machine virtualization to spin environments up and down.
With SDDC, it is possible for companies to replace their data center’s infrastructure with a virtualized environment and then deliver services and software as a service – Eric D. Brown
The question arrises, what technologies constitute and agile SDDC? What software should be leveraged to succeed at having an agile data center, a SDDC? The most important software to look at is software that addresses the top constraints in the data center. As the theory of constraints says, any improvement not made at the constraint is an illusion. So what are the top constraints in the data center. The top constraint, as found after working with 100s of companies and surveying 1000s of companies, is provisioning environments for development and QA. Why is that? It’s because almost every industry now is finding itself to be more and more a software industry from stock trading to booksellers to taxi companies to hotels. The competitive advantage is more and more about the software used to provide and sell the service. To build that software requires development and QA and thus development and QA environments. The hard part of an environment to provision is no longer the machine thanks to machine virtualization. The hardest part of the environment to provision is the data. Data that represents the production system is required to develop applications that use, display and manage that data. Data is typically kept in large complex databases such as Oracle, SQL Server, Sybase, Postgres and DB2.
Provisioning development and QA environments that rely on databases can be an expensive, slow endeavor. But like machine virtualization there is a new technology data virtualization that instead of making full physical copies, instead makes one copy of each unique data block on the source including a stream of changes blocks. With this “time flow” of unique blocks from the source database, data virtualization can provide copies in minutes not by actually making copies but by providing pointers back to the existing blocks. These existing blocks are read/writeable thanks to a technology of redirect on write which saves modified blocks in a different location than the original. It all sound a bit complex but when that’s the beauty of data virtualization solutions. They take the complexity, wrap it up into automated software stack and provide simple interface and APIs to provision full developer environments from the binaries, to the code files to the most complex and difficult part of the environment provisioning which is provisioning full running copies of the data. Included in most data virtualization solutions is masking as well since sensitive data is often required to be masked in development environments. The software defined data centers (SDDC) depend on machine virtualization and data virtualization.
What other technologies are also required?