Under the Hood: 4 Pillars of the Komprise Architecture


Revving up the Komprise Architecture with an Under the Hood Look at the Four Pillars

As the Co-Founders, we spent a great deal of effort investigating the problems that storage IT faces in the presence of data growth. In formulating the Komprise Intelligent Data Management solution, we determined the system should serve as a data substrate that sits behind the storage and stitches the hot and cold data storage together. With this concept we could then provide migration of data and transparent access to cold data, without being in the hot data or metadata paths or fronting expensive storage.

In this post, I will introduce the architectural pillars that drove how we built Komprise and show how those pillars helped satisfy the product requirements that we developed.

Distributed Scale-Out Architecture

The first architectural pillar is a distributed scale-out architecture, allowing Komprise to scale with data and storage. This pillar led to the grid of virtual appliances that run the data substrate and manage the storage.

The grid of virtual appliances acts in a plug-and-play scaling model, so proofs of concept can become production and small deployments can become large deployments, all by simply adding more virtual machines. From terabytes to petabytes and even exabytes, from dozens of shares to thousands of shares, we have customers who have scaled their installations without re-architected deployments, specialized hardware, or centralized bottlenecks.

Storage Agnostic & Non-Intrusive

The second architectural pillar is to be storage agnostic and non-intrusive. Being storage agnostic means Komprise can work with a heterogeneous collection of storage systems, exactly the sort of storage layout that most needs the management and stitching of disparate storage together. By being non-intrusive, Komprise can cooperate with a variety of systems and layouts, of both data and storage, without imposing restrictions on deployment or storage architecture.

Being storage agnostic, Komprise manages NFS, SMB, and object storage. It interacts through standard protocols to give the file-oriented view IT desires, while not requiring agents specific to each platform. Storage agnostic also means that Komprise does not compete with storage: Komprise partners with storage.

Being nonintrusive means fitting into and managing data migration within any storage environment without impacting the architecture or performance. This leads to an all-software solution where customers can run proofs of concept on production storage and data without deployment changes around their critical systems.

To see how you can simply connect Komprise to NetApp, Isilon, Windows File Servers and other NAS environments, watch the 5-Minute Demo.

Redundant, Transparent Hierarchical Filesystem

Another critical pillar underlying Komprise architecture is a redundant, transparent hierarchical filesystem. Without this pillar Komprise could not provide transparent access to data moved to secondary storage.

In effect, Komprise paints the NFS or SMB from the primary storage over whatever secondary storage Komprise has moved or copied the data to, whether it is object store, cloud, or another file server. The movement is transparent, meaning that the original namespace continues to provide access to the data that is now sitting on secondary storage.

Komprise is also a redundant filesystem in that you can access data from the secondary storage via Komprise through NFS or SMB protocols. To learn more, read our post Under the Hood: The Komprise Filesystem

Distributed Search & Analytics

The final architectural pillar is distributed search and analytics. In providing analysis at a shared level, Komprise sees and analyzes every file. Retaining the file analytics and allowing search and filtering, storage IT gains deeper insights and greater control over their data.

Deep Analytics, currently in beta, distributes a file-analytics database within the same grid of virtual appliances, thus retaining the easy plug-and-play scaling that is found in the rest of the Komprise Intelligent Data Management solution.

To learn more about the Komprise architecture, join us for a live webinar: Under the Hood: Four Pillars of the Komprise Architecture — August 2nd at 10am PT.

Stay tuned for our next blog, where we will review the implications of these architectural decisions on actual customer deployments.

Getting Started with Komprise:

Contact | Data Assessment