The modern data center has to deal with different use cases ranging from applications running inside of virtual machines or containers to massive amounts of unstructured data created by both users and devices. This data diversity almost always forces the organization to use multiple storage solutions, one for each data type and IO requirement. The problem is the cost of acquiring and managing these diverse storage systems inhibits the organization’s ability to modernize.
The Need for Consolidation
The value of consolidation spans more than an organization’s desire to reduce costs and simplify operations. Another key motivation for consolidation is to create a single pool or repository that all applications can access. A single pool of storage enables an application to integrate with data produced by another application.
Data lakes were supposed to be the answer to the need for a common storage pool. The problem is they became yet another silo of storage. Applications would need to copy their data into the data lake before other applications could leverage that data into its calculations. The separate data lake silo means that IT has to manually copy data to the data lake on a regular basis or create scripts to do so. As a result, the applications wanting to integrate data in from other applications as part of their process often never have the most recent copy of data.
IT needs a storage solution that can consolidate all or at least the large majority of its data storage into a single storage system that has enough performance to run production applications while at the same time is cost-effective enough to be a capacity storage area. Quobyte believes it is delivering on this need.
Quobyte is a parallel distributed file system that supports a variety of technologies like OpenStack, containers (Docker/Kubernetes), big data (Hadoop, Spark) and legacy bare metal (NFS, SMB, S3 and block). The company focuses on markets that are challenged by data growth and demand high performance like financial services, life sciences, chip designers, media and entertainment, as well as manufacturing (IoT).
The solution provides horizontal storage and organizations within those markets should be able to use Quobyte for the majority of their data requirements. It is delivered as software, and customers provide their own commodity hardware as storage.
It also provides replication and erasure coding for data protection. The protection is very granular, down to the file level. Data on the same volume can have different data protection policies applied to it. It also provides end-to-end checksums and can route around failed hardware.
From a performance perspective, each node in the cluster can support flash, hard disk drives or both. Placement of data is defined by policies based on file metadata and can change on the fly. The solution provides consistent low latency of less than one millisecond and can be optimized for high throughput or high IOPS workload requirements.
Finally, the system is designed to scale. Node counts can range from four to thousands of servers with a linear increase in performance and no measurable increase in latency. The hardware used for the nodes can be heterogeneous which provides the organization flexibility as the system scales.
In Quobyte’s latest iteration of the file system, it adds several key features including volume mirroring for disaster recovery, further simplification of the install and operational experience and native multi-protocol access.
The volume mirroring feature protects data across regions and clouds via continuous, asynchronous mirroring. As with single site data protection, DR protection can be assigned on a per-file basis. The software also translates between erasure coding and replication. If a data set is erasure coded within the data center, it is replicated to the remote site and if policy dictates re-erasure coded there.
Version 2.0 of the file system focuses on lowering the amount of Linux knowledge required for an installation, a capability that should make Quobyte more appealing to the traditional enterprise. The company claims IT can install the software and have it ready to receive data within ten minutes.
Finally, and potentially most important is version 2.0’s capability to share data across platforms and interfaces. This means data can be written from one platform like Windows or Linux and then read from another like Hadoop via S3.
Most storage systems require a trade-off. They either provide high performance or high scalability. Most can’t do it all, and if they do try the solutions become very expensive very quickly. Quobyte’s ability to mix media types, manage data movement and provide very granular data protection allows IT to solve a lot of problems with a single solution.
Add 2.0’s ability to provide shared access across platforms, and it becomes a solution that many industries could take advantage of. For organizations looking to modernize their data center and still support legacy applications without buying a half-dozen storage systems, Quobyte deserves strong consideration.