Vendors have started to provide storage products that support both file and object data to meet the demands of modern workloads.
By offering the two formats, a single storage system can handle both traditional and modern workloads, saving the expense and complexity of implementing multiple systems. This can be especially valuable to organizations that want to modernize their applications in manageable stages, rather than having to transition all their workloads at the same time.
These products differ, however, and organizations should understand their differences before they choose a system to accommodate their workloads. Dell EMC, NetApp and Pure Storage offer products that incorporate both file and object storage. Here, we explore how these products deploy both file and object storage, as well as the issues to consider when evaluating them.
Dell EMC PowerScale OneFS
Dell EMC PowerScale is a family of NAS products made up of Isilon and PowerScale nodes. The nodes support all-flash, hybrid and all-HDD configurations, offering a flexible and scalable infrastructure that can handle a wide range of unstructured workloads. Customers can combine different node types into a single cluster to meet varying application requirements.
The OneFS operating system runs on each node in a PowerScale cluster. OneFS simplifies storage and data management at scale and provides a software-defined architecture to store, manage, protect, secure and analyze data. The OS orchestrates the cluster components and delivers a unified storage pool to consolidate data and eliminate silos. OneFS also includes features such as snapshots, auto-balancing, disaster recovery and other data protections.
OneFS supports multiple industry-standard protocols that enable access to file-based storage, including NFS, SMB, HTTP, File Transfer Protocol and Hadoop Distributed File System. Starting with OneFS v9.0, the OS also supports the Amazon S3 protocol, which makes it possible for users to access a cluster's file-based data as objects. Modern applications that use object storage can take advantage of the file-based system, without admins having to deploy additional components.
OneFS incorporates S3 as a first-class protocol that supports bucket and object operations. An organization can ingest content and metadata through the protocol, which is implemented over HTTP or HTTPS. The platform's file protocols can then access the content, just like they would access regular file data. OneFS maps each S3 object to a file and each S3 bucket to a directory under an access zone base path. The S3 protocol is implemented on top of the file service engine along with the other storage protocols. This makes it possible for both file-based and object-based applications to access a single file system on the same storage platform.
NetApp OnTap
NetApp OnTap is a data management infrastructure product originally designed to handle NAS and SAN storage systems. NetApp added support for S3 object storage with the release of OnTap 9.8. OnTap software provides a common set of features that operates across deployment architectures, which simplifies data management for hybrid clouds.
An organization can use OnTap to create an integrated, scalable storage environment made up of NetApp's All Flash FAS and FAS hybrid-flash storage systems, as well as its FlexPod converged infrastructure. NetApp offers the OnTap Select virtual appliance to deliver software-defined storage on commodity hardware. Cloud Volumes OnTap manages and controls data on AWS, Microsoft Azure and Google Cloud. Organizations can move data between these architectures to optimize performance and cost efficiency.
If decision-makers want a product that supports both file and object storage, they must understand how the vendor implements those formats.
The OnTap platform supports all major industry-standard protocols, including NFS, SMB/Common Internet File System, iSCSI, Fibre Channel and, now, S3. With the addition of S3, users have a single data management platform that can handle small-scale object workloads, along with file-based and block-based applications. That said, the S3 implementation does not support all OnTap features or S3 functionality. For example, it doesn't support Cloud Volumes OnTap or S3 object versioning.
When OnTap is used for object storage, the FlexGroup volume is the architecture that underlies an S3 bucket. A FlexGroup volume is a scale-out container made up of multiple constituent member volumes that are managed as a single volume. The FlexGroup volume provides a single namespace to work with the data. The S3 protocol is typically implemented over HTTPS (rather than HTTP), enabling clients to connect to FlexGroup volumes that contain S3 buckets. However, they can't use the NAS or SAN protocols to connect to these volumes, limiting data access to object-based applications.
Pure Storage FlashBlade
FlashBlade is an all-flash storage platform that unifies file and object storage. Pure Storage designed the platform to deliver high performance and density in a compact package that simplifies management and provides elastic scale-out capabilities. Each FlashBlade chassis can house up to 15 blades and deliver up to 15 GBps bandwidth, while supporting tens of billions of files and objects.
Purity for FlashBlade (Purity//FB) software powers the FlashBlade platform. Purity//FB scale-out storage management includes features such as file replication, object replication and file system rollback, which enables fast file system recovery from snapshots. The Purity//FB platform includes a massively distributed transactional database. The database is built on a modern key-value architecture that enables FlashBlade to distribute both data and metadata across storage nodes.
Each blade in a FlashBlade system joins raw NAND flash with an Intel Xeon system-on-chip processor. Pure Storage has rearchitected the SSD's flash translation layer to eliminate bottlenecks. It has reprovisioned the DRAM to improve parallelism. Each blade comes with file and object system software that provides data and resiliency features, and it includes an embedded software-defined switch fabric. FlashBlade provides a single namespace that can scale from 199 TB to 7.8 PB.
FlashBlade is designed with a core object store that supports NFS and S3 access. The distributed transactional database enables file and object access without the use of gateways, which can affect performance and efficiency. Using a specialized object naming convention, the database eliminates the need to separate file and object systems. It also facilitates a native scale-out architecture that can distribute metadata, files and objects. FlashBlade provides full compatibility with S3 APIs. However, object and file storage are implemented in separate namespaces, with no support for cross-protocol access.
The products mentioned above from Dell EMC, NetApp and Pure Storage demonstrate some of the key factors to consider. For example, not every product supports all S3 functionality or offers cross-protocol access to the data. FlashBlade might provide full compatibility with the S3 APIs, but only PowerScale OneFS offers cross-protocol access. In the end, it will come down to which features an organization requires and which storage product can best support its workloads now and in the future.
