From Dell: Why Automated Discovery Changes Game for NVMe/TCP?

By Ihab Tarazi, CTO and SVP, infrastructure solutions group core, Dell Technologies, Inc.

Back in August, I wrote about how you can boost your workload performance with an NVMe IP SAN and how Dell Technologies, Inc. helps simplify adoption by validating the components used within the NVMe/TCP ecosystem and providing automation with SmartFabric Storage software (SFSS). In this post, I want to focus on the automated discovery to simplify the deployment.

SFSS and NVMe/TCP ecosystem

When deploying NVMe/TCP, organizations often face a common challenge: managing connectivity between the host and storage array. This is due to the considerable amount of work involved, including manually configuring each host to point at the appropriate array-based Discovery Controller, monitoring the established connections and remediating any connectivity failures that may occur.

How SFSS’s automated discovery solves challenge?
SFSS removes the need for configuring each host interface to each storage interface one-by-one. Instead, host and subsystems automatically discover the SFSS instance and register with it. It then acts as a broker to help establish communication between the hosts and subsystems. Customers can configure the host and subsystem relationships via zoning, just as with FC.

Ease of deployment, especially in medium and large enterprise environments, is a principal reason some users prefer FC over other IP-based technologies, despite the cost-saving potential and performance improvement. For network infrastructure deployment, Ethernet costs is up to 89% less than FC in some solutions.⁽¹⁾

In addition, a standards-based discovery service such as SFSS, that is scalable and automated, is a boost for other operating environments (e.g., edge and cloud).

How Automated Discovery works?
The existing discovery approach for NVMe/TCP requires customers to repeat the following three steps on all hosts for each subsystem. First, the host admin connects the host to a Discovery Controller at a specific IP address. Next, the storage admin provisions Namespaces (storage volumes) to the host NQN. Only then can the host admin now discover and connect to the I/O controllers on that subsystem.

This process might work well in smaller fabrics with a few dozen host and storage endpoints. However, manually configuring and discovering each storage subsystem on a host quickly becomes difficult and error-prone for administrators as fabrics scale.

With the new, standards-based, Centralized Discovery that includes support for automated discovery, after the initial setup, users only need to repeat steps 2 and 3 below for each host on each subsystem. This cuts the deployment time while minimizing potential errors.

1. Host and subsystems automatically discover the CDC, connect to it, and register discovery info

2. Zoning performed on CDC (optional)

3. Storage admin provisions namespaces to the host NQN. Storage may send zoning info to CDC

4. After zoning, the host receives AEN, uses get log page, and connects to each I/O Controller

How Automated Discovery works

What’s new for December 2022 release of Dell NVMe IP SAN?
Dell continues its commitment to simplifying its customers’ ability to benefit from NVMe/TCP. The December release expanded the validated ecosystem to PowerStore running PowerStoreOS 3.X. PowerEdge running VMware ESXi 8.0 and PowerSwitch Z9664.

Complete list of validated Dell’s products

As for deployment automation, SFSS now supports VMware ESXi 8.0 and RHEL 9.1 (Tech Preview), with a complete list of OS support below.

Complete list of SFSS’s OS support

Dell is excited about its progress and committed to helping customers improve workload performance while saving costs with NVMe/TCP Technology. Read NVMe IP SAN solution brief and SFSS spec sheet to learn more.

⁽¹⁾ Based on Dell internal research calculation (September 2022). Actual results may vary.