Zero Overselling was Netz0's founding principle before it was a policy: every resource we sell is a resource we actually have, reserved for the customer who bought it. This page explains what that means and lists the specific commitments behind it. The narrative sections describe our philosophy and practices; the numbered commitments in "The Commitments" section are incorporated by reference into the Netz0 Terms of Service and bind us contractually.
Why "Zero Limit" instead of "unlimited"
You will see "Zero Limit" on our plans where we impose no numerical cap on a specific resource. Zero Limit does not mean infinite — it means we add no artificial limit, while physics still applies. Unlimited email accounts on a mail server still share the server's finite storage and CPU; the honest statement is that the count is uncapped and the underlying resources are the real constraint, disclosed on the plan.
We avoid the word "unlimited" because the hosting industry spent two decades abusing it. Nothing in a datacenter is unlimited: no one has invented the infinite hard drive, which is why the largest cloud providers meter everything and charge for actual usage. Plans advertising unlimited everything for a few dollars were collecting payment for resources that could never be delivered, and the industry has largely — though not entirely — grown out of it. Where a resource genuinely has no metering, the honest term is unmetered, not unlimited, and that is the term we use.
The modern version of the trick is subtler: the hidden sub-limit. A server advertised with a 1 Gbps port that quietly caps packets per month. "Unlimited storage" with an undisclosed file-count (inode) limit that exhausts at a few gigabytes. Advertised speeds conditioned on unpublished fair-use thresholds. The headline number is technically true; the buried limit makes it meaningless. Our first commitment below exists specifically to make that trick impossible here.
What you pay for is what you get. That has been the deal since day one, and it is why our plans have real, published numbers instead of fantasy ones.
Don't just buy hosting — buy a piece of the internet.
The Commitments
The following commitments are incorporated by reference into the Netz0 Terms of Service and apply to all Netz0-hosted services.
- No hidden sub-limits. Every limit that constrains a resource you purchase is published on the plan page and in your account. Advertised bandwidth carries no undisclosed packet-rate caps. Advertised storage carries no undisclosed inode or file-count limits. Advertised speeds are not conditioned on unpublished fair-use thresholds. If a limit is not published, it does not exist.
- Storage is pre-allocated. All disk space sold — virtual, cloud, or dedicated — is reserved before you buy it and fully available at any time. You may use none of it or 100% of it; it is yours either way.
- Bandwidth and transfer are reserved. Bandwidth packages and monthly transfer allotments are pre-assigned to the network at their maximum possible consumption, and your contracted transfer is available in full — never withheld or throttled below what you purchased. You may consume your allotment as fast as your port speed allows; once consumed, additional capacity requires an upgrade unless previously arranged. Advertised port speeds are full-duplex and dedicated. On virtual and cloud plans, network speed is engineered to deliver the full purchased rate at least 95% of the time and no less than 80% of it the remainder, as measured by Netz0's network monitoring.
- A hard cap on accounts per server. Shared platform servers (Cloud Sites) host no more than 100 customer accounts per physical server, regardless of how lightly those accounts use their resources. One account may hold multiple domains and websites without affecting this count. We monitor load continuously and rebalance accounts when needed so every customer receives the full resources purchased.
- On shared plans, only compute is shared. Shared plans carry fixed maximum allocations for CPU, RAM, and disk I/O. Storage, bandwidth, and network speed remain fully dedicated even on shared plans and may run at their maximum 24/7. Because servers are not oversold, we can allow accounts to peak into idle capacity at our discretion; accounts that sustain maximum compute continuously belong on — and will be guided toward — a cloud or dedicated plan.
- Dedicated means dedicated. Dedicated plans assign full processor cores and fixed RAM that no other customer can touch, configured to match or exceed equivalent bare-metal performance. You may run them at maximum continuously. Failed hardware is replaced at no cost; obsolete hardware is replaced with equal or better, including free migration of your data.
- Cloud plans state their model honestly. Cloud instances receive the fixed allocations of their plan, placed on the next available cluster; underlying hardware may change and no specific machine is guaranteed. Compute performance depends on cluster load — we maintain excess capacity and rebalance clusters to keep it high, but sharing compute is the nature (and the price advantage) of cloud plans. Storage and network commitments 2 and 3 apply in full. For cloud services without separate bandwidth packages, we reserve at least 25% network headroom and can expand capacity within 24 hours.
- Business-grade hardware only. All hosted services run on rack-mounted, enterprise-grade server hardware built for continuous datacenter operation — server platforms from established manufacturers (such as Supermicro or Dell), server-class processors (such as Intel Xeon or AMD EPYC), ECC registered memory, and drives rated for datacenter use. We do not host customer services on consumer PCs or desktop-class parts. Network routing and switching under our control uses proven professional equipment (vendors such as Cisco or MikroTik, among others). If an enterprise-grade part is temporarily unavailable, we may use the best available high-performance alternative until a proper replacement is installed — disclosed practice, not hidden practice. All hardware is tested and benchmarked before deployment and monitored continuously, with components replaced or repaired proactively.
- The network is never oversold. We connect through at least three Tier-1 network providers and sell network capacity in slices of what we actually have. Every non-cloud plan's bandwidth is pre-allocated even when idle. The network is engineered for continuous availability; our real-world track record has been effectively 100% for years, but we do not advertise 100% — nothing in this industry truly is — so our contractual commitment is the 99.99% defined, with credits, in our Service Level Agreement.
- Proper facilities, restricted access. Services run from certified datacenter facilities meeting our standards for power redundancy, cooling, physical security, and network capacity, with access restricted to authorized personnel. All paid services include our SLA from day one.
Where the numbers live
Plan-specific allocations — storage, transfer, port speed, compute — are published on each plan page and in your account. The uptime commitments and credits are in the Service Level Agreement. If you ever find a limit affecting your service that is not published, report it: under commitment 1, that is a bug in our documentation, not a term of your service.