NVMe/TCP: A Step Towards the Cloud Operating Model

With Purity//FA 6.4.2, Pure leads the way again, being the first vendor to offer NVMe-oF for all the major fabric networks—FC, RoCE, and TCP.

Cloud Operating Model

image_pdfimage_print

Whether for processing the data generated by your business or keeping that deluge of data protected, boosting storage performance is always a priority. With its parallel data streams and low-latency access, NVMe was designed to capitalize on flash’s groundbreaking performance and get around SCSI’s inherent bottlenecks.

Now, with Purity//FA 6.4.2, FlashArray supports NVMe/TCP, the final missing piece required for mainstream adoption of NVMe over Fabrics, in addition to FC and RoCE. Since NVMe/TCP runs on the ubiquitous Ethernet network, businesses get end-to-end performance on an easy-to-deploy and manage network. What’s more, since Ethernet is the network of choice for the cloud, NVMe/TCP enables seamless hybrid cloud deployments and paves the path for cloud operating models such as Pure Fusion.

With NVMe/TCP, organizations now get all the benefits of an NVMe fabric without committing to a dedicated storage network like Fibre Channel or adding specialized RoCE equipment to their Ethernet network. NVMe/TCP delivers the same advantages as NVMe/FC and NVMe/RoCE: increased throughput and decreased latency, and benefiting from using Transport Control Protocol (TCP), which offers:

  • Scalability: TCP supports widely available, easy-to-expand networks for growing needs.
  • Ease of use: TCP is a “simple” protocol with common and well-known standards that IT staff are already familiar with and trained to manage.
  • Reliability: TCP has always been one of the most reliable Ethernet transports, with features such as packet acknowledgment, packet loss recovery, congestion control, etc.
  • Worldwide usage: TCP/IP is widely deployed and especially prevalent in cloud and hybrid environments. 

35% Better Performance than iSCSI for Demanding Workloads 

NVMe/TCP gets around the bottlenecks that made iSCSI less than ideal for performance intensive applications, and unlike iSCSI, NVMe protocol was designed for flash media instead of spinning media. Why is this important? Modern-day applications with strict service level agreements drive the need for the low-latency, high-performance, on-demand networks that NVMe-oF can deliver. In particular, applications that use microservices, everything-as-a-service, or another cloud or hybrid cloud model will benefit from a more efficient storage protocol that runs on Ethernet.  

To help customers make the rapid decision to adopt NVMe/TCP, we had a clear goal: to provide significant performance improvements, so they see instant results. The performance advantage of NVMe/TCP over iSCSI was clear in recent benchmark testing. Testing on the FlashArray//X 90R3, comparing NVMe/TCP results to iSCSI results, we saw up to 35% improvement in latency and a 25% increase in throughput/IOPS with FlashArray//X under a heavy load. Squeezing out additional performance in an already existing network shows why we think NVMe/TCP will increase in popularity rapidly.

And unlike some of our competitors, the benefits of NVMe/TCP are available across our full FlashArray product line, not just our performance-oriented FlashArray//XL and FlashArray//X R3 systems. So even if you’re using FlashArray//C R3 for storing snapshots or running less-demanding workloads, you’ll be able to use NVMe/TCP and see benefits. 

To dive deeper into NVMe/TCP for FlashArray, including testing results, jump into the NVMe/TCP deep-dive blog.

Get NVMe/TCP Today through Pure1 Self-service Upgrades

With the release of Purity//FA 6.4.2, all FlashArray R3 customers can now get NVMe/TCP. Pure continues to deliver non-disruptive updates, meaning no downtime and no performance impact on users or applications. And for FlashArray systems running Purity//FA 6.3.1 or beyond, you can initiate a self-service upgrade through Pure1

Pure1 self-service upgrades will guide you through the entire upgrade process. This includes running multiple health checks throughout the upgrade, ensuring the correct binaries are installed, and performing the upgrade as if you had support on the line. The entire upgrade is performed within Pure1 and can be executed anytime.

flasharray

Read our detailed how-to blog for self-service upgrades in Pure1.

Want to be always up to date with the latest in Purity//FA?

This Purity//FA 6.4.2 announcement is the latest in our commitment to a regular monthly release cadence. To help you keep track and stay organized, we’ve built a Purity//FA 6.4 Release Highlights roundup page. There you’ll find links to blogs, announcements, demo videos, podcasts, and the all-important release notes. 

To get this and more in Purity//FA 6.4.2, simply upgrade your FlashArray fleet on your schedule through the self-service upgrade capability in Pure1.

Integrating NVMe/TCP with Hybrid Cloud Solutions

As organizations increasingly adopt hybrid cloud architectures, the role of advanced storage protocols like NVMe over TCP becomes even more critical. Pure Storage’s hybrid cloud solutions leverage NVMe/TCP to provide seamless, high-performance connectivity across on-premises and cloud environments, ensuring consistent data access and management.

Key Benefits of NVMe/TCP in Hybrid Cloud:

  1. Seamless Integration Across Environments: NVMe/TCP offers robust compatibility with existing Ethernet infrastructures, enabling organizations to extend high-performance storage solutions across hybrid cloud setups without the need for specialized hardware or complex configurations.
  2. Enhanced Performance: With up to 35% faster data access compared to traditional iSCSI protocols, NVMe/TCP ensures low-latency and high-throughput storage performance across both on-premises and cloud environments. This is particularly beneficial for latency-sensitive workloads such as analytics, artificial intelligence, and machine learning.
  3. Scalability and Flexibility: NVMe/TCP provides a scalable solution for organizations expanding their storage capabilities to the cloud. Whether you’re running containerized workloads in Kubernetes or virtual machines in VMware, NVMe/TCP delivers consistent performance for dynamic workloads.
  4. Simplified Management: Pure Storage’s hybrid cloud solutions offer unified management tools that streamline operations across on-premises and cloud deployments. With native support for NVMe/TCP, administrators can easily optimize storage performance while reducing complexity.
  5. Data Mobility and Resilience: NVMe/TCP enhances data mobility by enabling efficient replication and migration between environments, supporting disaster recovery and business continuity in hybrid cloud architectures.

Use Case: Hybrid Cloud Analytics

With NVMe/TCP, organizations can process data locally for performance-sensitive operations and seamlessly transfer it to the cloud for long-term storage, analysis, or compliance. This hybrid model ensures that critical operations benefit from low latency, while the cloud provides cost-effective scalability.

By integrating NVMe/TCP into your hybrid cloud strategy, you can achieve unparalleled performance and flexibility, positioning your organization to handle the evolving demands of modern applications and workloads. Pure Storage continues to lead the way in delivering innovative solutions that empower businesses to maximize their hybrid cloud investments.

Modern Hybrid Cloud Solutions