A file storage architecture for enterprise performance at scale

[StorageExperts]

To achieve enterprise performance at scale, file storage must have an architecture capable of breaking through performance bottlenecks. HPE GreenLake for File Storage has an architecture that does just that.

–By David Yu, HPE Storage Product Marketing

Enterprise performance at scale is a must for modern, data-intensive applications such as artificial intelligence (AI), machine learning (ML), Big Data, deep learning (DL), high-performance computing (HPC), large data lakes, and high-speed access to data archives. These applications span a wide range of industry verticals, including life sciences, financial analytics, and media and entertainment and process enormous volumes of file data.

To achieve performance that scales, file storage must have an architecture that breaks through performance bottlenecks. HPE GreenLake for File Storage has just such an architecture.

Enterprise performance at scale

I wrote a blog when we announced HPE GreenLake for File Storage that addressed the need for enterprise performance at scale. I detailed just what we mean when we talk about enterprise performance at scale – and where current systems fall short.

As an analogy, I described two different vehicles. One, a racecar that’s a complex, highly customized system driven by a solitary, highly skilled driver. The car can set a record by reaching a breakthrough speed for a few seconds, but it has no practical, day-to-day transportation value. On the other hand, a vehicle that is also fast but can easily transport many people is much more effective, even if it doesn’t set a land speed record. In this analogy, the passengers represent the data that’s our real concern. Enterprise performance at scale isn’t simply flat-out performance that reaches an unprecedented peak, at an instant in time, for a small data set. Instead, it’s fast, sustained performance that spans the entire scale of your data and supports workloads that process extremely high volumes of data.

Another way to understand enterprise performance at scale is to imagine a sprinter who can run a marathon at a world-class 100-meter pace. It’s a sustained, high-speed performance over a long stretch. And this is just what HPE GreenLake for File Storage delivers. Where legacy NAS infrastructure hits a limit and can’t keep going, HPE GreenLake for File Storage continues to scale and sustain performance even as it processes huge amounts of data.

An architecture for enterprise performance at scale

What allows HPE GreenLake for File Storage to achieve enterprise performance at scale? It’s a solution designed for exabyte scale that has two key components: VAST Data software with a DASE^TM (Disaggregated Shared Everything) architecture and HPE Alletra Storage MP modular, resilient hardware providing for independent scaling of performance and capacity.

Under the hood, HPE Alletra Storage MP compute nodes are connected to HPE Alletra Storage MP storage nodes over an NVMe fabric. The storage nodes contain ultra-efficient, all-NVMe storage for blazing-fast performance, and every compute node can access all storage nodes.

HPE GreenLake for File Storage eliminates all the drawbacks of legacy, scale-out NAS, including crosstalk, rebuilds, and interdependencies that only increase geometrically with cluster size. As the architecture is designed to grow to exabyte capacities, it future proofs your storage investment. HPE GreenLake for File Storage delivers linear performance scaling while overhead remains flat. The result is more scaling of performance than most organizations can ever consume. And independent scaling of performance and capacity provides flexibility and efficiency in meeting exactly the performance and capacity points you need.

All of this means you can easily scale up and scale out. Add compute and storage resources as needed, instead of getting another box or appliance that sacrifices flexible and efficient scaling. The architecture of HPE GreenLake for File Storage eliminates migration issues and is intended to give you an infinite data lifecycle.

Even better, with HPE GreenLake for File Storage, there is no controller resource contention. Controller nodes are stateless, as all the metadata is kept in the storage nodes. With persistent, fast, Storage Class Memory, all nodes have access to the entire data structure. You have a global namespace for your storage system for data sharing, and you can have up to N-1 controllers fail and still keep going. There are no rebuilds on server failure, and for data protection, a new erasure code algorithm can rebuild a failed SSD without reading the entire very large stripe that the system uses. This results in a very low overhead of around 2.7%.

Shared nothing vs. shared everything

All the above is not possible with shared-nothing systems. Have a look at the contrasting summaries of the two different architectures below.

With shared nothing, you get:

• Ownership of partial data
• Fast if directly attached to the servicing node, but at scale can cause contention and incurs significant overhead
• Node failure can slow performance down significantly

With shared everything, you get:

• All compute nodes have access to all data
• Scale performance without adding overhead from East-West node traffic
• Up to N-1 failures allowed for compute nodes (you just need one functional node)
• Load balancing without resource contention

HPE GreenLake for File Storage: Supercharge data-intensive applications

Combine all this goodness with the HPE GreenLake edge-to-cloud platform and you have a truly compelling file storage solution. In addition to blazing fast performance, you get: an intuitive cloud experience plus over 50 other cloud services spanning storage, networking, compute, and workloads – not to mention the security, control, performance, and cost advantages of on-prem infrastructure. And there’s more: You get streamlined deployment, easy file share creation, unified storage management with a single cloud console, and automated, nondisruptive upgrades. With read performance that is 80x faster than the NFS capabilities of legacy NAS and 100s of GB/sec of throughput and beyond,¹ you have a file storage architecture designed for the enterprise performance at scale that is essential for today’s modern workloads

HPE GreenLake for File Storage is the ideal file storage solution for today and tomorrow, as you continue to amass, process, and extract value from data. So seize your competitive advantage by supercharging your most demanding, data-intensive applications with the enterprise performance at scale of HPE GreenLake for File Storage.

Read the first blog in this HPE GreenLake for File Storage series:
A cloud management experience for file storage minus security risks or trade-offs

And watch for a third blog exploring in detail how this solution enhances productivity and efficiency with faster time to insights, high ROI from capital investments, and radically simple file data management.  

To learn more

Read Top 12 reasons to choose HPE GreenLake for File Storage

Watch the HPE GreenLake for File Storage demo

^{1 HPE Storage Substantiation}

Meet Storage Experts blogger David Yu, HPE Storage Product Marketing

David has a key product marketing role in HPE’s storage business, covering such areas as file-and-object storage, scale-out storage, cloud-native data infrastructure, and associated cloud data services.

Storage Experts
Hewlett Packard Enterprise

twitter.com/HPE_Storage
linkedin.com/showcase/hpestorage/
hpe.com/storage