Real Time Analytics Requires Real Time Data Processing
Real time data requirements are increasing and becoming the norm whether you’re driving a Tesla with the latest version of software or tracking cyber threats. New and existing enterprise workloads are requiring higher application-level performance than current flash technologies can often deliver. Response times and low latency are key attributes for an agile enterprise, and the volume of data collected by organizations is growing exponentially, thanks to key trends that are driven by the following use cases:
• AI (artificial intelligence), ML (machine learning), and DL (deep learning)
• Real-time analytics
• IoT (Internet of Things, requiring orders of magnitude more data)
• Video, social, mobile, and blockchain
Next Generation Storage Class Memory
These use cases require memory hungry applications with massive data sets to be analyzed and shared in real time, driving the need for more new approaches to data management that deliver radically lower latency. Next generation memory technologies like Persistent Memory (PMEM) are designed for this purpose. What is PMEM? Think of it this way: Persistent memory is accessed like volatile memory (RAM) by the CPU and application, achieving orders of magnitude lower latency than direct attached storage, but it retains its contents after a power loss like a regular storage device.
Real Time Use Cases
As mentioned above, performance-sensitive databases (relational and NoSQL), real-time applications and in-memory databases are the primary workloads. The most relevant use cases would be fraud and cyber threat detection, financial trading and real-time market analytics, health care diagnostics, forecasting, social media environments leveraging personalization algorithms, Internet of Things (IoT) workloads, or AI/ML-driven inference models. Any environment where customers need ultra-low latency as measured at the application-level, the ability to support huge data sets, or want to enable enterprise-class data services for these demanding tier 0 applications will benefit from PMEM.
Challenges and Solutions
Regular memory, or Random-accessed memory (RAM) is directly addressed by the CPU. PMEM is byte addressable, so an application accessing the data in PMEM needs to be rewritten to access this data and use it. NetApp MAX Data software simplifies the adoption of PMEM by enabling applications like Oracle to access data on PMEM without any changes. This means customers looking to adopt this next generation storage can do so immediately. No waiting for vendors to modify their code for PMEM.
What is MAX Data?
MAX Data (Memory Accelerated Data) software runs on the application server and provides a file system that spans PMEM and a storage tier. Applications stored on this filesystem get instant access to the data for both reads and writes (which means MAX Data is not caching software for those unclear on the concept who shall go nameless). With MAX Data, you get vastly improved application performance with high throughput and ultra low latency. However, performance is just the beginning…
Advanced Data Services
In the past real time applications were restricted to volatile memory. While PMEM brings write support to real time workloads, it also brings the need for enterprise data services. As you explore using persistent memory, you’ll need enterprise data services; MAX Data includes the ability to mirror and protect persistent memory within a server as well as utilize snapshots for fast data recovery. With NetApp, you can tier data to an ONTAP-based AFF all-flash system and leverage all the data management capabilities in ONTAP including high availability, cloning, backup and disaster recovery.
Next Generation PMEM Eco-system is Getting Ready for MAX Data
We are very excited about MAX Data, and not just because of capabilities the product brings to our customers around performance, data protection, and efficiency. MAX Data represents NetApp as a leader in next generation flash/PMEM ecosystem. MAX data already has the consideration of large, global enterprises for real time data analytics capabilities. We’re also excited to partner with Intel, Cisco, and Lenovo as part of the broader ecosystem of vendors looking to enable the adoption of PMEM in the server.
Don’t Look Back
The NetApp Data Fabric is built for the future, supporting both traditional and emerging applications, such as NoSQL databases and artificial intelligence. It offers the industry’s only unified data management platform that supports SAN and NAS, all-flash storage, software-defined storage, hybrid cloud, and cloud. You can scale up and out dynamically in seconds or in minutes, instead of taking hours or days. And you can allocate applications to where they run best, whether it’s on the premises or in the cloud. With MAX Data, you can extend the data fabric capability all the way into your servers with your applications and data that are critical to your business.