Inside GreenLake for File Storage, Technical Overview from HPE

By Iris Wagner, senior software enginer, HPE Storage

Today’s digital landscape is seeing the rise of big data and AI. Now more than ever, organizations need file storage solutions that can keep up with ever-growing file storage demands – and still deliver high-performance, scalable capacity without breaking the bank.

The company offers a response to those challenges with the HPE GreenLake for File Storage. Teaming up with Vast Data Ltd., the firm is changing the storage industry. The next-Gen Alletra MP Storage platform will be running Vast Data’s software stack. This disaggregated scale-out unstructured data solution will give businesses the advantages to be leaps and bounds ahead of the rest because of your ability to tap into the insights from your data.

What sets GreenLake for Storage apart from the rest?
Consider these 4 major differentiators:

1. Alletra MP Storage hardware with its ability to scale performance and capacity independently, as the common hardware platform for both the file solution as well as the block solution separately
2. The revolutionary disaggregated and shared-everything architecture
3. Effective data reduction technologies
4. Strategic advantages from its integration into the GreenLake edge-to-cloud platform ecosystem

Now for a deeper dive.

Alletra Storage MP appliance

Alletra Storage MP and disaggregated architecture
Alletra Storage MP hardware is designed to move away from using the purpose-built hardware platform uniquely designed for each storage solution. Instead, the new platform provides a single common, disaggregated hardware platform for all the different HPE SDS solutions. The installed software on the common hardware defines the personality of the storage array. For example, configuring a switched multi-node system with a software stack for file workloads gives you Alletra Storage MP for File. Composing from the same exact hardware components in different combinations with a block-based software stack gives you Alletra Storage MP for block storage. The disaggregated storage architecture separates compute nodes, JBOFs, and switches into network-connected components. This design makes it possible to scale performance and capacity independently to fit any growing needs.

For the initial release of GreenLake for File Storage, the company is providing the option to configure it anywhere from the smallest configuration to up to the maximum configurations in 2 different aspects. The 1st is the maximum usable capacity configuration with 2 controller nodes and 12 JBOF enclosures with 3.3PB capacity. The 2nd is the maximum performance configuration of 8 controller nodes and 6 JBOF enclosures with IO/s capabilities over 1 million. Shortly after the initial release, by the end of the calendar year, look for larger configuration options available with 30TB drive support and 12PB total capacity.

The software stack adopts the shared-everything architecture, which runs on top of the disaggregated hardware architecture. Traditional unstructured data solutions typically have a shared-nothing architecture which was never meant to service the performance required at the scale of modern-day explosion of data. On the contrary, the software architecture of this solution was born to scale.

Thanks to storage class memory providing persistent storage with memory-like speed, all controller nodes can have the same fast access to the metadata of the entire data structure without incurring the overhead of east-west ‘chatting’ between nodes. This architecture not only alleviates any resource contention within the controller nodes caused by drive or enclosure ownership, but also provides a single global namespace for the entire storage space. Controller node failures or additions cause zero disruption to hosts and will not trigger a re-indexing of the entire file system that slows down performance significantly, which is common with other unstructured data solutions. The controller nodes of Alletra Storage MP hardware are designed to be stateless in this architecture and thus any controller node failure won’t cause any disruption in servicing IO/s to the clients.

Click to enlarge

Data reduction technologies
For a business to be able to scale to exabytes of capacity of flash, it is critical to have data reduction capabilities offered by the GreenLake for File Storage solution. Leveraging Vast Data’s data reduction technologies, the storage solution is designed to achieve high reduction ratios with real-life workload data, such as HPC, life science data, animation data, etc. Data reduction methods includes global de-dupe that looks for duplicated data blocks from the entire existing data set. This de-dupe process can be far more effective than local de-dupe, because the shared everything architecture gives all controller nodes access to the entire data set without the tradeoff of east-west traffic to take away CPU cycles. Here’s another example: Instead of finding an exact match for a block to de-dupe, the similarity reduction algorithm finds blocks that are highly similar but with small differences. It then compresses them together and stores only a reference block and a small set of byte-level deltas.

Part of GreenLake edge-to-cloud platform ecosystem
This solution complements the existing GreenLake Data Services Cloud Console services in the GreenLake platform eco-system for managing your HPE infrastructure and workloads such as block, file, and backup, and recovery all in one place.

Data Services Cloud Console provides simple file data management with an intuitive cloud experience along with the capabilities of managing your other HPE infrastructure and workloads. It eliminates management complexity by removing infrastructure management silos. It provides a cloud-managed single console accessible from anywhere and from any device. Within Data Services Cloud Console, 2 services support file service management: Data Ops Manager enables to manage your organization’s Alletra Storage MP infrastructure for file and block. The File Storage application is where consumers can create, protect, and manage their file storage.

In addition to being managed from the GreenLake platform, an on-premises UI is also available with the same feature sets, providing a simple, intuitive alternative to cloud-based control.

Accelerate your most demanding, data-intensive workloads with enterprise performance at scale
I’m excited about the benefits and opportunities that the GreenLake for File Storage solution will bring to your organization. It has never been easier, faster, and more reliable to store files – and unleash the power of your data to give your business the insights you need to sharpen your competitive advantage.

Resources:
HPE GreenLake for File Storage solution brief
Top 12 reasons to choose HPE GreenLake for File Storage
Blog: HPE transforms data lifecycle management with expanded HPE Alletra portfolio with new file, block and data protection services  
Blog: HPE GreenLake for File Storage: Scale-out file storage to supercharge data-intensive apps
Video: HPE GreenLake for File Storage – 3-minute demo 
Video:HPE GreenLake for File Storage – technical demo 
Video:HPE GreenLake for File Storage – explainer video
Video:HPE GreenLake for File Storage – chalk talk