Comparing vSAN ESA to VergeFS

vSAN ESA supports only NVMe SSDs. A year ago that was an acceptable trade, but times have changed. The cost of flash capacity has more than tripled in the last year, with 15.36 TB drives now costing more than $10,000. VCF 9.1 adds QLC support, but restricts it to cyber recovery clusters, so production capacity stays on TLC NVMe. ESA also lacks hard disk drive support, and HDDs now make more sense for workloads that aren’t performance-sensitive, like file servers and lightly used databases. You can buy a 24 TB production-quality hard drive for less than $1,200.

VergeFS requires NVMe drives for metadata, but you can place all other data on drives of almost any type, including SAS and SATA SSDs, QLC SSDs, and hard drives. VergeOS can also live migrate VMs between these tiers. IT can change its data placement strategy in seconds and switch back if performance problems appear. The key is that IT stays in control rather than relying on an auto-tiering algorithm, and auto-tiering caused hybrid storage to fail once before.

vSAN ESA deduplicates after the write is already committed. Data sits at full size until post-processing reaches it, so you have to purchase the extra capacity up front. ESA deduplication is also a storage feature, and RAM and the network get no benefit. Finally, deduplication isn’t available for clusters smaller than three nodes. Which makes it less than ideal for small remote data centers and edge locations.

VergeIO’s global inline deduplication is integrated into the core of VergeOS. It benefits every aspect of the data center, not just VergeFS. There is also no minimum node count for deduplication to be active.

VergeOS’s global deduplication also supports many-to-one replication, which is common in branch office deployments like Topgolf’s. vSAN ESA can send delta changes to a recovery site, and VMware documents no check for whether that site already holds a data block sent from another location.

VergeOS uses its virtual data center (VDC) technology to simplify disaster recovery. A VDC encapsulates an entire workload: all of its VMs and data, all of its network configuration, and all of its storage settings and policies. The whole VDC replicates and fails over as one unit.

VCF builds protection groups from sets of VMs. It can’t protect tenants and virtual data center objects. Networks must be in place at the recovery site before replication starts, making network configuration protection a separate project. NSX configuration travels to the recovery site by backup, as a separate job in NSX Manager.

Licensing adds to the gap. vSAN ESA needs a per-VM Advanced Cyber Compliance or SRM subscription for remote replication, plus a vSAN ESA target. VergeOS includes replication in its core license at no extra charge.

In a VCF deployment, vSAN ESA requires the VCF management stack to be in place before storage is operational. That stack includes VMs for vCenter, a three-node NSX Manager cluster, SDDC Manager, and VCF Operations. The customer installs and owns them, makes them redundant, and maintains a separate DR strategy.

VergeOS is a single operating system. Everything installs at once, so storage and networking are ready as soon as you install VergeOS. There are no infrastructure VMs. The first VM you create is yours.

A customer with hardware three or more years old will more than likely need to replace some, if not all, of their servers to run vSAN ESA. Storage is the biggest issue. ESA rules out SAS and SATA drives entirely, and many servers from that era were built around them. VCF 9 has also put several recent CPU generations, including Intel Cascade Lake, in deprecated mode ahead of removal.

VergeIO has never deprecated hardware. VergeOS supports almost any server built in the last ten years, along with a variety of media types and interfaces. The efficiency of a single-code-base infrastructure OS lets customers run their existing hardware longer, with more VMs per physical host. Our customer Formica said in a recent case study that VergeOS extended the life of its European data center servers to 15 years, keeping them in service until 2034.