Hyperconverged infrastructure (HCI) brings compute, storage, networking, and backup into one software layer. A handful of standard servers becomes your own private cloud. Scope of services covers planning, deployment, migration, training, and annual maintenance for both the VergeIO and Arcfra product lines.
The four points below are compiled from publicly reported market news (source: market reports), not our own statistics. Actual terms are governed by your quote and the vendor's contract.
The buy-once-and-pay-only-maintenance model has ended. Licenses must now be renewed when they expire.
Pricing is based on CPU core count, and each host has a minimum purchase quantity. Even a lightly configured host must be licensed up to the threshold.
Modules that used to be sold separately are now bundled into suites, so you pay for features you don't use.
The VMware Cloud Service Provider (VCSP) program is ending. Customers who bought through a service provider under this program need another path.
The question isn't whether VMware is good software. It's what your data center budget will look like for the next three to five years if you sign this quote. Related reading: What to do about VMware license price increases
A traditional data center has a three-tier architecture: the machine that computes (the server), the machine that stores (an external storage array), plus the storage area network (SAN, usually built from Fibre Channel switches) that connects the two. Three types of equipment, three vendors, three management consoles. Hyperconverged infrastructure (HCI) writes all three functions into one software layer, installed on a handful of standard x86 servers:
One piece of software turns a few ordinary servers into your own private cloud. The unit of expansion shrinks from "a whole set of equipment" to "one server."
The difference isn't just equipment count. Shrinking the unit of expansion from "a whole set of equipment" to "one server" changes how you budget.
We support both product lines. The difference is which scenario fits.
VergeOS installs on standard x86 servers, and once running, each host acts as a virtualization host, storage node, network node, and backup node at the same time. Licensing is calculated "per physical server" - the price is the same regardless of how many CPUs or cores you install, which is the most direct cost difference compared with per-core pricing. Notable features are VDC (Virtual Data Center - a fully virtualized data center that can be cloned and nested, at no extra charge according to the vendor) and ioProtect. The latter leaves production untouched and only requires licensing the disaster-recovery site, letting you validate the platform at minimal risk. The vendor is a US company.
Arcfra Enterprise Cloud Platform (AECP) integrates virtualization, distributed storage, software-defined networking, Kubernetes (a container cluster management tool), backup and disaster recovery, and multi-cluster management into one platform. Licensing is a per-CPU-core subscription, with no extra charge for storage capacity or node count - a clear advantage for customers whose data volume grows quickly. AKE lets you build Kubernetes clusters through a graphical interface (built in from the Standard edition up), paired with the Neutree model-serving platform for on-premises AI inference. The vendor, ARCFRA PTE. LTD., is registered in Singapore.
Beyond scale and features, one point in vendor selection is often overlooked and should be settled first: whether your industry or your customers require supply-chain source review. Government agencies, some financial institutions, and Taiwan subsidiaries of foreign companies may have restrictions on the source of ICT products or internal procurement rules. This directly affects which platform fits. Please confirm the applicable rules with your compliance or procurement department. We adjust the recommendation once that's confirmed, and can help request country-of-origin declarations and compliance documents through the supply channel where needed. See How to choose hyperconverged infrastructure: three evaluation criteria. We source vendor products and support through authorized channels in Taiwan, and provide planning, deployment, migration, training, and annual maintenance services.
Hyperconverged software still runs on physical servers in the end. Node count and CPU core configuration directly determine licensing cost and headroom to expand. Where a project needs it, we plan GIGABYTE high-performance and GPU servers alongside the software, finalising both in the same architecture design - so you don't end up with software and hardware from separate vendors pointing fingers at each other when something goes wrong.
On-premises AI: your data never leaves your own data center. Adding GPU servers to the same cluster keeps both models and data inside the company. This matters most directly for industries handling trade secrets, personal data, or confidentiality obligations. Arcfra's Neutree model-serving platform (an open-source edition plus an enterprise edition) and VergeIO's VergeIQ, released with VergeOS 26 (announced October 2025), are both newer modules. We recommend a proof of concept (POC) to confirm performance and compatibility before deploying to production.
A hyperconverged platform's built-in console only sees its own cluster. But your company's IT also has databases, application systems, network equipment, hundreds of employee PCs, and batch jobs that need to finish every night. This section covers the highlights only - see the Hitachi JP1 IT operations management platform for the full picture.
JP1 manages the machines. ERP manages the business.
Whether a server is down or a batch job finished is JP1's job. For orders, inventory, and receivables/payables, see ERP Business Management.
HULFT is often mistaken for a Hitachi product, but its actual vendor is Saison Technology Co., Ltd. (TSE Standard: 9640). It moves data securely and reliably between different systems; JP1/AJS3 decides when to move it and how to handle failures. Technically, AJS3 can schedule and monitor HULFT jobs - the actual integration is confirmed case by case for each environment. Integration with other vendors' equipment is mainly via standard syslog, SNMP trap, or API forwarding; the exact scope of support varies by product and version.
Five stages. The first two leave production completely untouched - you decide whether to proceed once the numbers are in.
Both platforms run on standard x86 servers. Changing software doesn't scrap your hardware. We recommend testing the VM export and migration path during the proof-of-concept (POC) stage, and writing acceptance criteria into the contract for an exit path. You can also start with a minimal commitment: VergeIO's ioProtect only requires licensing the disaster-recovery site, and Arcfra can take over an existing ESXi environment first and replace it gradually.
Our engineers handle first-line intake and resolution during business hours. When vendor assistance is needed, we file the case. We source vendor products and support through authorized channels in Taiwan; the actual scope of service follows your annual maintenance contract.
No, they operate at different levels. The hyperconverged console only sees its own cluster; JP1 sits above it, pulling company-wide alerts into one screen and then handling scheduling and asset inventory. The two are layered, not substitutes for each other.
No. A common first step is the disaster-recovery site or a test environment, which minimizes downtime impact while still being a real-world test. When and whether to migrate production fully is decided once the POC numbers are in.
As a renewal deadline approaches, the most expensive decision is "no time to evaluate, just sign." Complete the three items below before you sign.
Leave your contact details and a description of your needs, and an engineer will follow up to help assess architecture planning, deployment approach, and the scope of annual maintenance.