Why horizontal scaling is important？

 　　What is vertical scaling?

　　Scaling up means adding resources to a single node in a system. Like adding a hard drive to a PC, it has been a common way for enterprises to perform storage upgrades over the past few decades. Over time, however, this approach has revealed more and more limitations.

　　For storage systems, because vertical expansion is simply adding hard drives or flash drives to an existing architecture to increase storage capacity without adding CPUs and memory to help the entire system handle more capacity delivered to the host. This means that when storage capacity is increased, storage performance tends to degrade.

　　

　　For example, if the current array processing performance has reached a bottleneck, expanding capacity at this point will affect the overall performance of the LUNs previously mapped to the host, as more LUNs compete for system resources that have already reached a bottleneck. In turn, this can affect backup and recovery times and other mission-critical processes.

　　What is horizontal scaling?

　　Scaling out is the process of replacing or adding new hardware to an existing IT system. As capacity is scaled out, performance increases linearly with capacity. Because each node that scales out has a separate CPU, separate memory, etc., the performance of the entire cluster does not degrade with the increase in capacity after scaling out space, but rather increases.

　　

　　The horizontal scaling design is suitable for unstructured data, where the data can be distributed across multiple nodes to improve resiliency and performance. Typically, with this type of data, I/O profiles do not require block-level deterministic latency (as is the case with block-based I/O). This is why horizontal scaling and object storage solutions stand out in their design. A horizontally scalable solution also allows individual volumes to have more capacity than a single node, so object storage or file systems that need to support large capacities are well suited for horizontally scalable designs.

　　Essentially, horizontal scaling allows the shortcomings of previous vertical scaling to be addressed, with core benefits including

　　- Freedom from the capacity and performance constraints that limited older devices.

　　- Reducing the cost of complex infrastructures to quickly benefit from newer architectures and disk drive densities without the need for expensive forklift upgrades.

　　- Better hardware simplifying system management, promoting redundancy and improving uptime more easily.

　　- Making it relatively easy for organizations to physically scale up in the future. The complexity of traditional vertically scaled architectures can create the risk of business disruption when upgrading, while horizontal scaling is relatively much easier.

　　In addition, before horizontal scaling became popular, organizations often purchased storage systems that were much larger than the required capacity (e.g., RAID solutions required over-purchasing disks for parity and hot spares) to ensure there was enough disk space available for future expansion. However, if this expansion never occurs, or if demand does not reach the expected level, much of the initially purchased disk space is wasted.

　　With a horizontally scalable architecture, the problem of high initial investment costs is avoided, and if storage demand increases beyond expected levels, new arrays can be added as needed, with essentially no limits.

　　

　　Vertical scaling is not obsolete from now on

　　Data explosion today, a strong scalability is a must for any IT environment, since horizontal expansion to make up for the many shortcomings of vertical expansion, vertical expansion will be eliminated? Actually not, vertical scaling still has a place.

　　For small deployments, a vertically scalable architecture can meet all the requirements of an application. For deployments larger than a petabyte (or in cases where a large single namespace is required), a better choice is definitely horizontal scaling. A horizontally scalable architecture introduces better resiliency at scale by using erasure coding rather than RAID protection. A horizontally scalable design also provides greater efficiency in a geographically distributed model than a full replication of the entire device.

　　In the long run, however, IDC expects that 80% of data will be unstructured by 2025, and Gartner data also shows that unstructured data capacity is expected to triple from 2019 to 2024. As you can see, the vast amount of unstructured data will place greater demands on enterprises to use and manage and store it in the coming decades, while also posing unprecedented challenges to the performance and scalability of storage infrastructure.

　　In this regard, Dell EaseUS PowerScale (formerly Isilon), the industry's premier scale-out NAS storage, enables customers to truly "defy the challenge, defy the odds" when faced with the challenge of unstructured data innovation.

　　

　　*Powered by Intel® Xeon® processors with software-defined infrastructure and agile cloud architecture, PowerScale delivers superior performance and efficiency to accelerate demanding file workloads, enabling enterprises to leverage the value of their data capital and accelerate the digital transformation of their business.

　　Simply put, the benefits of PowerScale are reflected in three key areas▼

　　Scalability and Simplicity

　　PowerScale leverages OneFS' powerful scale-out architecture capabilities to allow customers to dynamically configure the right amount of capacity and performance they need, as needed, without over-provisioning storage or performing push-back upgrades.

　　More importantly, with a single namespace, single file system environment and enterprise-class data services benefits, customers gain ease of use, flexibility and high performance through increased efficiency and new automation capabilities with the OneFS operating system.

　　

　　Any data, any location

　　Any Data, Anywhere means that PowerScale supports up to 8 protocols including NFS, SMB, HDFS, S3, REST and more, while providing access to any client and supporting any user's needs. Providing unlimited flexibility for enterprise run workloads that can both store data using one protocol and access it using another.

　　Anywhere means that PowerScale can bring OneFS to more locations by providing the capability of new PowerEdge-based all-flash and NVMe nodes, truly making OneFS "everywhere" in the storage environment, whether in the data center, edge, core or cloud, with a unified OneFS operating system that can manage storage in the same way.

　　

　　Massive data, deep insights

　　In providing intelligent insight, PowerScale gains insight into storage infrastructure and unstructured data with the addition of CloudIQ and DataIQ.

　　CloudIQ makes it easier to monitor the operational status of the entire data center system, while DataIQ enables customers to easily find and utilize data in file and object storage, enabling unified "management" of unstructured data stored in all kinds of unstructured storage in the enterprise (not limited to all unstructured storage in Dell EMC, such as Isilon, ECS and PowerStore storage devices).

　　

　　This makes it more convenient and easier for customers to migrate and archive data, and more efficient to do analysis and utilization based on these data in the future, truly allowing customers to get maximum value from unstructured data and turn data into insight.