The Cohesity Software Architecture
The Hidden Cost of Trying to Make Legacy Solutions Work
Traditional storage architectures haven’t kept pace with the demands of explosive data growth and the need to support new workflows such as big data analytics. They have created silos of management, systems and workflows, leading to massive data sprawl and countless copies of data. Over the years, organizations have tried to address new business imperatives by adding, swapping, and integrating new solutions into legacy architectures. All these stopgap measures and purpose built solutions have created complex, isolated and disjointed processes for managing Data Protection, Test/Dev, File Services and Analytics workflows.
Cohesity was founded with the core principle of building the next generation, truly distributed general purpose storage system that eliminates the fragmentation in data storage and puts an end to the decades-long “Band-Aid effect” that has plagued data storage solutions.
Cohesity has built a web-scale, distributed, multi-layered architecture that comprises of physical and software layers working together to support the application layer that ultimately manifests the many different storage workflows.
It is physically a shared-nothing distributed architecture. It allows us to take advantage of low-cost, high performance commodity hardware. Each node has compute and storage resources connected by a dual 10GbE network. Intelligent software coordinates these nodes to operate as a single, coherent system. There is no single source of bottleneck allowing for infinite and complete linear scalability.
Base Layer: Cohesity Open Architecture for Scalable Intelligent Storage (OASIS) Filesystem
The Cohesity Data Platform is built on the Open Architecture for Scalable Intelligent Storage (OASIS), the only file system that combines infinite scalability with an open architecture flexibility that consolidates multiple data storage workloads onto a single platform.
OASIS was built from the ground up to be a fully distributed, strongly consistent and versatile storage system. A key part of the file system is our unique snapshotting technology called SnapTree. SnapTree allows for frequent and near instant snapshots, while keeping the data fully hydrated. This supports the most stringent RPO/RTO goals imaginable. Another key part is a true global deduplication capability that ensures that the same dedup block is not written twice across the cluster.
OASIS is itself comprised of several components, each one handling a key function to provide a completely fault tolerant, infinitely scalable architecture. These components enable seamless scaling of the system as new nodes are added. This also ensures high availability of all hardware and software components. They also abstract hardware resources including compute and different tiers of storage (SSDs, HDDs, Cloud) to manage Quality of Service levels for different workloads that co-exist on the system.
The full power of the OASIS file system is unleashed through a set of interfaces that together constitute the service layer. This layer is key to unleashing the power of the filesystem in numerous different storage workflows. The services layer supports storage protocols such as NFS and SMB. It enables replication between different clusters to support disaster recovery and data availability. It has built-in search and a MapReduce framework to support instant search and ad-hoc file content analytics.
The application layer supports all the storage workflows which currently include comprehensive data protection, Test/Dev, File Shares, Analytics and Cloud integration. The application layer provides cloning, scheduling, policy management, backup software, application adapters, integration with LDAP/AD, data archival, pre-built analytics apps, as well as a powerful framework for creating custom analytic applications. The storage workflows are controlled through a single pane of glass built on top of RESTful APIs.
To dive deeper, download the Cohesity Architecture White Paper.