What criteria should we use to evaluate virtualization vendors?

When evaluating virtualization vendors, consider factors such as virtualization capabilities, management and automation features, cloud compatibility, performance and reliability, security and compliance, cost and licensing, and support and ecosystem.

How do licensing and support models differ?

Licensing models vary between vendors, with some offering per-socket, per-VM, or subscription-based models. Support models also differ, with some vendors offering 24/7 global support while others may have more limited coverage.

What SLAs should we demand?

Demand SLAs that align with your organization's uptime and performance requirements. Consider factors such as response times, resolution times, and availability guarantees.

How does virtualization impact security and compliance requirements?

Virtualization can impact security and compliance by introducing new attack surfaces and data management challenges. Look for vendors that offer robust security features, such as encryption, microsegmentation, and compliance certifications.

Fibre Channel or NVMe over TCP?

By Allan Waters, Director Product Marketing, Nutanix

April 2, 2026 3:08 pm |

min

As enterprise IT teams modernize their infrastructure, many are reevaluating long‑standing assumptions about storage networking. NVMe over TCP (NVMe/TCP) brings the high‑performance NVMe protocol to standard Ethernet networks, eliminating the need for dedicated Fibre Channel fabrics while delivering latency and throughput that rival traditional SANs. By leveraging widely available Ethernet infrastructure, NVMe/TCP provides real opportunities to reduce cost, simplify operations, and align with how modern data centers are already architected. With Nutanix now integrating with external arrays over NVMe/TCP, customers gain a practical, future‑ready alternative to Fibre Channel, one that supports cloud‑centric architectures, scales effortlessly, and accelerates the journey to a more efficient hybrid environment.

Attribute	NVMe over TCP (Ethernet)	Fibre Channel (FC)
Performance (latency/IOPS)	Very low latency; comparable in many workloads with proper tuning	Industry standard for predictable low latency
Cost (CapEx/OpEx)	Uses commodity NICs/switches; lower hardware and optics cost	Higher-cost HBAs, FC switches, optics; specialized skills
Operational complexity	Leverages existing IP tooling and staff skills; easy to scale	Requires SAN expertise, separate management plane
Ecosystem & future roadmap	Rapid ecosystem growth; standards-based NVMe/TCP spec	Mature roadmap (Gen7/8) but more specialized evolution
Security & features	TLS, DIGEST options; runs over IP networks	Built-in SAN isolation and mature zoning features
Best fit	General purpose, disaggregated, cloud-native, Nutanix + external arrays	Apps needing near zero packet loss

Quick guide: key considerations and decision points

Existing skillset: Do ops teams know Ethernet/IP better than SAN? NVMe/TCP reduces training and operational friction.
Budget & scale: Do you want lower optics/HBA spend and easier scale‑out? NVMe/TCP runs on commodity Ethernet gear.
Vendor support: Is your HCI vendor supporting external NVMe/TCP arrays? Nutanix has announced external storage integrations (Everpure, Dell PowerStore) over NVMe/TCP.

Why NVMe/TCP is compelling

Standards-based and production-ready. The NVMe/TCP transport is an official NVMe specification with features like optional inline integrity and TLS, enabling enterprise deployments over standard IP networks.
Performance parity in practice. Independent tests and vendor studies show NVMe/TCP can match or closely approach FC performance for many real-world workloads when networks and host stacks are optimized. Ceph and other scale systems report strong NVMe/TCP scaling at datacenter scaleCeph.
Lower TCO and simpler ops. Ethernet NICs, switches, and optics are broadly cheaper and more flexible than FC HBAs and directors; operationally teams can reuse existing IP tooling and skills, reducing OpEx.
Easier adoption with Nutanix + external arrays. Nutanix’s move to support external NVMe/TCP arrays (and partnerships like Everpure) means customers can keep existing storage investments while modernizing compute and management on Nutanix Cloud Platform. Yes, a FC Pure array can be converted to TCP/IP.

Risks, trade-offs, and mitigations

Network design matters. NVMe/TCP requires careful Ethernet design (loss, congestion control, QoS) to match FC predictability; mitigate with dedicated VLANs, QoS, and monitoring.
Maturity for extreme determinism. For ultra‑low jitter, some shops may still prefer FC; evaluate with proof‑of‑concepts and workload benchmarking.
Operational change management. Plan training, runbooks, and phased migration to avoid disruption.

Recommendation & next steps

If you’re considering NVMe/TCP as part of your Nutanix strategy, now’s the perfect time to explore further. Nutanix Cloud Infrastructure (NCI) with external storage provides a powerful and flexible path forward, and you can learn more on the NCI with external storage webpages for detailed architecture guidance, validated platforms, and deployment best practices.

For deeper technical insights into migration, be sure to check out Bhavik’s NVMe/TCP migration blog, which walks through practical steps and real‑world considerations.

Taken together, these resources will help you build confidence, validate design choices, and plan a smooth transition toward a modern NVMe/TCP-powered Nutanix with external storage environment.

©2026 Nutanix, Inc. All rights reserved. Nutanix, the Nutanix logo and all Nutanix product and service names mentioned are registered trademarks or trademarks of Nutanix, Inc. in the United States and other countries. All other brand names mentioned are for identification purposes only and may be the trademarks of their respective holder(s)