I’m certain anyone reading this is well aware of the statistics about Data growth and how it is impacting storage requirements across all industries. This isn’t a new challenge in our industry, but the conversation does have an added twist when we consider the impact of IoT. We commonly read about companies experiencing anywhere from 20% to 50% year over year growth. The terms “exploding!”, “explosive”, and “exponential” are usually found in articles associated with data growth (and now I’ve used all three in one post). While this data growth continues to be spurred on by the traditional sources associated with business data (databases, documents, email, etc.), we are seeing even greater capacity requirements being generated by IoT devices.
For this post, when I speak of data from IoT, I am lumping together data generated by security video, temperature gauges, vibration sensors, connected cars… you get the idea. In fact, according to some sources, IoT data is 10x the growth of that associated with traditional data sets. And, IDC estimates IoT devices will grow to 28.1 Billion by 2020. So, data collected from these devices, and storage solutions needed to maintain this data, will become increasingly important.
Among our clients, we see a tremendous growth in the need for storage to maintain Security Surveillance Video. Beyond simply providing a place for video streams to be written, our clients are analyzing the video; utilizing software to perform anomaly detection, facial recognition, etc. I shared a couple posts recently, written by two of my colleagues at Zunesis, that expands on this topic. And, Analytics isn’t isolated to video only. The value of IoT devices is that they capture data at the edge, where it is happening, and this is true across all IoT devices. Once collected, software can perform analysis of the data to derive meaning beyond the data points and, in many cases, produce actionable insights. So, storage required for IoT data needs to be able to hit large scale quickly and have performance characteristics that allow analytics in near real time. And, of course, this storage still needs to provide reliability and availability associated with any business-critical data.
To meet the storage requirements defined above, HPE has created a hardware platform and partnered with two software defined storage (SDS) companies to provide solutions for scale-out storage that will grow from a couple hundred terabytes to petabytes and provide both the reliability and performance required of the data generated by the ever-expanding number of IoT devices. The HPE hardware is part of the HPE Apollo family. The software that utilizes this hardware comes from Software Defined Storage providers, Qumulo and Scality. Here is a summary for each of these solution components:
The Apollo Family of systems from HPE are each designed to provide compute, storage, and networking that meet the needs of both scale-up and scale-out requirements. They are targeted at workloads supporting Big Data, analytics, object storage and high-performance computing.
The scale-out compute part of the HPE Apollo System portfolio includes the Apollo 2000 System for hyperscale and general-purpose scale-out computing, the Apollo 4000 System Family is targeted at Big Data analytics and object storage while the Apollo 6000 and 8000 Systems are designed to support HPC and supercomputing. Density, ease of management (all incorporate HPE iLO management), and efficient rack-scalability are features shared by all members of the portfolio.
Qumulo is a software defined scale-out NAS that scales to billions of files in a flash-first design. With the Apollo/Scality solution, you can scale from 200TB to over 5PB of usable capacity. This solution uses advanced block-level erasure coding and up-to-the minute analytics for actionable data management. The file services provided by Qumulo are also supported in the public cloud, currently on Amazon Web Services.
Use cases include:
Scality is a a software defined Scalable Object Storage solution that supports trillions of objects in a single namespace. With the Apollo/Qumulo solution, you can scale to over 5PB of usable capacity. The access and storage layers can be scaled independently to thousands of nodes that can be accessed directly and concurrently with no added latency.
Use cases include:
So, yes, data footprint is growing and won’t be slowing down anytime soon. If your data set is outside the traditional business data sets and requires scale-out storage that supports large numbers of files and the ability to perform actionable analysis quickly, then you probably need to look outside of the traditional scale-up storage solutions and look at solutions purpose-built for these large-scale workloads. HPE Apollo, Qumulo, and Scality are a great starting point for your research.