When we set our sights on the data management space, we wanted to take an approach that fit with our core principles. Most of the incumbents in the space were complex, they locked data in proprietary formats, and they had significant issues at scale in terms of capacity and performance. The Komprise solution needed to be simple, resilient and scalable to meet the challenges of enterprise data management and the explosion of unstructured data.
To analyze, tier, move, and provide non-disruptive access to petabytes of data and billions of files you need an elastic scale-out approach. The Komprise Elastic Grid is a distributed, shared-nothing, scale-out software architecture that was designed from the ground up to manage data at massive scale.
The Members of the Komprise Elastic Grid are:
- Komprise Observers, which are software virtual appliances installed close to storage devices and connected via a cooperative, distributed algorithm enabling scalability with load balancing and high availability.
- The Komprise Director, which manages the Observers and runs as a cloud software service or on-premises.
The Komprise Grid provides the underlying scalable layer upon which the Komprise Intelligent Data Management solution is built. This ensures that all analytics and management functionality provided by the solution is scalable and highly available.
The Komprise Intelligent Data Management Solution Provides the Following Core Functionality:
- Analytics to provide insight into your data and enable planning data management functions and capacity planning.
- Data management functions, which include migration, cloud tiering, and replication/cloud DR.
- Deep analytics that enables you to run custom queries to search, tag and manage your data at a file level in order to create a Global File Index.
The Grid dynamically expands and shrinks to enable high availability, scalability, load balancing and failover. The Komprise Grid runs adaptively in the background so there is no noticeable performance impact of running Komprise in an environment. If your environment is overloaded during the day, Komprise will adaptively slow down the analysis rate and automatically speed it up at night when the load subsides.
Key functions of the Komprise Elastic Grid:
Key architectural principles of the Komprise Elastic Grid:
How the Komprise Elastic Grid Works
The Grid scales both performance (needed to move data fast) and capacity (needed to manage billions of files and petabytes of data) by parallelizing work both within and across Observers with the ability to elastically add more Observers on the fly. Deployed as standard VMs, the Observers are added to the Grid through the Komprise Director user interface. The more nodes added the greater the scale; shares are load balanced across the Observers in the Grid.
What Makes the Komprise Elastic Grid Unique
The Komprise Elastic Grid balances operations on the many shares managed by the system across the Observers in the Grid. As Observers are added or removed, the Grid automatically rebalances the load of the shares and the operations to achieve a high degree of parallelism and scale in analysis and transfer. In cases where even more parallelism is needed, e.g., to migrate a large share, an Observer will engage other Observers to assist.
The Komprise Elastic Grid is stateless and can be readily brought back to life even if all the Observers have failed. Each Observer maintains its high-level configuration (e.g., which shares it is handling) with the Director and periodically updates it. If an Observer goes down, a new Observer communicates with other Observers in the Grid to ensure it has the latest state before becoming a productive member of the Grid. This statelessness enables simpler recovery and enables us to create a large grid of Observers.
Even though different shares are mapped to different Observers, the end users see Komprise as one logical file system. That is, the grid of Observers is not visible to the end user. Komprise provides a “Komprise Access Address” (KAA) that is used to mount Komprise as a file system. The KAA is used to transparently access any data that has been tiered by Komprise from any share. Komprise ensures that any access request coming in to the KAA is properly routed to the Observer that is managing the share from which the request originated. If that Observer is down, Komprise will automatically forward that request to another Observer which has been selected to manage that share. The Komprise Grid uses a watchdog mechanism to continuously ensure the KAA is hosted on a healthy Observer.
Conclusion: Faster, Smarter, Proven Data Management
Each customer we work with has a different data management challenge depending on industry, growth rate, infrastructure, geography, and so on. The Komprise Intelligent Data Management solution is a flexible service giving enterprise IT organizations the ability to first get a holistic picture of their storage, backup and cloud data and then to take charge. By controlling their unstructured data, they get control of their IT budget and can make the optimal choices to accelerate their business rather than simply expanding legacy storage.
Komprise allows customers to start out small and grow as their data grows by simply adding more Observers. The shared-nothing aspect of the solution allows our customers to expand the solution without restriction.