Software-defined secondary storage platform: That would be a mouthful if spoken, so let me explain what we mean by it and why we focused on building this architecture rather than relying on an appliance model. Software-defined signifies that the Imanis Data software stack does not depend on storage servers based on a storage area network (SAN) or network attached storage (NAS). It also means that our software stack is not tightly coupled to an exact hardware configuration. Early on, we decided to rely on commodity servers to run our software stack in lieu of our designing an appliance. This decision provides several advantages for customers.
- The current generation of commodity servers from Intel is quite powerful in terms of CPU, memory, and storage; our customers use those servers to run their production Big Data deployments such as Hadoop, Couchbase, or Cassandra. This type of hardware is also perfectly adequate for their data management needs. By using similar servers from the same vendors, our customers can affordably homogenize their data centers for two huge functions: production Big Data deployments and Imanis Data-based data management.
- A software-defined solution also lets customers start off with a small number of servers. The Imanis Data platform can be grown incrementally by adding more commodity servers and does not require “forklift” upgrades. Adding additional nodes to the Imanis Data cluster is a simple exercise, and the cluster is automatically rebalanced.
- The Imanis Data software stack is qualified with several different enterprise versions of Linux such as RHEL, Ubuntu, Centos, and Oracle Linux. The Imanis Data software can be installed on the Linux version that prevails in a company’s data center. Because our platform is OS-independent, generic data center issues—say, operating system installation or management and monitoring of physical hardware—can easily be done with powerful data center tools like Chef and Puppet. Essentially, management of the Imanis Data servers is no different from management of other hardware that hosts Big Data applications such as Hadoop or Cassandra.
- A big advantage of a software-defined stack is that it can easily be deployed either in the cloud or on-premises. Several of our customers run their Big Data applications in the cloud or plan to move to the cloud; other customers run such applications on-site on bare metal or in virtual machines. Cloud or premises, companies can flexibly deploy the Imanis Data software stack.
- Finally, from a business perspective, the biggest advantage is pricing. Customers win when they can source hardware independently of their software. The advantage of a software-defined platform is the convenience of using commodity configurations and not having to pay a hefty premium for an appliance. Thus, we can pass along to our customers the cost savings of a software-defined platform.