Blueprint for Choosing the Right NVMe Storage Solution

This blog post is an introduction to NVMe over Fabrics (NVMe-oF) and provides a blueprint for choosing the right storage solution. If you already know what you need and are looking for more information on the Lightbits Complete Data Platform, look no further.

Non-Volatile Memory express (NVMe) has transformed the storage industry since it emerged as the state-of-the-art protocol for high-performance solid-state drives (SSDs).

NVMe accesses flash storage with the help of the PCI Express or PCIe bus, which has thousands of parallel command queues. This makes NVMe faster than SSDs and hard disks with only a single command queue. While initially designed for high-performance direct-attached PCIe SSDs, NVMe was later expanded with NVMe over Fabrics (NVMe-oF) to support a rack-scale remote pool of SSDs.

What is NVMe over Fabrics (NVMe-oF)
NVMe-oF is an end-to-end standard that facilitates efficient data transport between servers and storage systems. NVMe-oF provides the best storage characteristics for cloud-native applications. It extends the low latency and high-performance characteristics of NVMe across networking fabrics, including

• NVMe-oF over Remote Direct Memory Access (RDMA) – Offers a way to exchange information between two computer systems across a network while offloading network processing from the main processor of either machine.
• NVMe-oF over Fiber Channel – Transfers data between storage arrays and servers using standard fiber channel (FC) Protocol, which supports access to shared NVMe flash.
• NVMe-oF over Transport Control Protocol/Internet Protocol (TCP/IP) – Uses the TCP transport protocol to transfer data across IP (Ethernet) networks.

These three enable the creation of an end-to-end NVMe storage solution with high performance and low latency.

NVME-oF in action
NVMe-oF ensures efficient CPU usage and faster connectivity between applications on servers and storage. Data centers that rely on direct-attached storage (DAS) experience further consolidation due to using it. Thanks to broader bandwidth availability and faster network speeds, apps can now be run on shared network storage. NVME-oF makes the storage architecture more parallel while eliminating bottlenecks in modern applications.

The industry has widely accepted that the NVMe-oF model will replace the Internet Small Computer Systems Interface (iSCSI) protocol as the communication standard between compute servers and storage servers and become the default protocol for disaggregated storage.

NVME-oF and Containerized Applications
Containerized applications are an example use case for NVMe-oF. Kubernetes is an open-source container orchestration system that automates software deployment, scaling, and management. A server running cloud-native applications might easily have Kubernetes running dozens or hundreds of microservices, all in need of efficient, parallel access to storage.

Therefore, cloud-native applications orchestrated by Kubernetes have distinctive storage needs that NVMe-oF can meet. Most Kubernetes apps are data intensive. This makes them the ideal candidate for high-performing storage delivered through NVMe-oF. And is particularly true for stateful applications like databases. These need low latency from storage to drive acceptable levels of application performance. Storage for such apps must also be able to scale easily if the app itself is to scale.

NVME-oF and the Public Cloud
When it comes to the public cloud, operators want the same enterprise features they enjoy on-premises. However, the economics of block storage, especially when considering high-performance volumes, is a challenge. Ease of use is not negotiable. High-performance databases and data analytics applications are just two use cases that require high-performance block storage in public clouds. This is the area in which NVME-oF excels.

For block storage, it boils down to the cost of capacity, performance, and performing snapshots and restoring data from them. The right NVME-oF environment can deliver the efficient use of storage resources. These encompass the likes of compression, thin provisioning, automatic cluster scaling, and no charge snapshots and restores. Using NVME-oF in this environment can deliver next-level volume performance with over 1.5M IOPS per volume, almost 6X more than existing solutions.

NVME-oF and Real-Time Data Analytics in Financial Services
In financial services, the speed benefits of NVME-oF cannot be ignored. Organizations are struggling with latency issues, especially when it comes to high-frequency trading platforms. Market data is time-sensitive, with the need to deliver real-time analysis a business imperative. Many companies in the sector are turning to NVME-oF to access an environment where they get faster, reliable storage while making real-time analysis possible.

NVME-oF and Complex Data Processing in the Telecom Industry
NVMe-oF also addresses the complex data processing requirements of the telecom industry. The past several years have seen telecommunication networks generating massive amounts of data as more people turn to mobile data for work and entertainment. This puts pressure on effectively dealing with this influx of data at the scale required to provide a competitive advantage.

Fast access to timely data is the future for any telco looking to remain relevant. Using NVMe-oF as the foundation, organizations in the telecom sector can access relevant information at a rate of magnitude faster than before.

NVME-oF and Private Clouds
While there are many cloud storage solutions on the market, some cloud providers and enterprises choose to  build their service on a private platform, or private cloud. There are many reasons why organizations choose to build a private cloud, including increased security, regulatory compliance, and more flexibility. One of the most important factors for private cloud storage is the speed of data access, which depends heavily on the performance of the backend storage devices. NVMe-oF can be a benefit here.

A modern private cloud deployment model typically includes a set of compute nodes, which are disaggregated from the NMVe storage nodes. Compute and storage nodes can be dynamically connected using fabric interconnects. The goal of NVMe-oF is to add no more than 10 microseconds of latency to the storage system, compared to a direct-connected NVMe SSD, making the difference between local storage and remote storage very small.

NVME-oF and Edge Clouds
NVMe-oF addresses the high speed and low latency requirements of edge clouds. By its definition and design, edge computing seeks to eliminate the need to move data from endpoints to a cloud and back again. 

There are a variety of NVMe-oF solutions out there today but the only one that allows for the separation of storage and compute is NVMe/TCP without requiring any changes to the client.

I won’t provide more detail on edge clouds here, for more information on NVMe-oF for edge clouds read these other blogs: Edge Cloud Explained – Top Benefits And Considerations and How Can NVMe Over TCP Improve Edge Storage? 

Advantages and Disadvantages Between Various NVMe-oF Protocols
If you have read so far, you might be wondering, which NVMe-oF is best for you.

Why choose Lightbits when it comes to NVMe-oF?
NVMe-oF over TCP/IP offers several benefits for cloud-native applications beyond great accessibility, lower overhead costs, and reduced complexity. It provides virtualized and centralized storage pools that can act as local flash storage. With this approach, NVMe-oF over TCP/IP means accelerated application performance, coupled with easy and efficient scaling.

Lightbits delivers the NVME-oF cloud-native and redundant storage environment essential to get an unmatched combination of enterprise-rich data services, resiliency, high performance, and scalability. This simplifies infrastructure management and operations for database and analytics applications.

Additional resources:

• Data Storage 101: Edge Cloud Explained – Top Benefits And Considerations