What “unlimited” and “unmetered” actually mean in hosting plans

No host gives you unlimited anything. The limits are real, they are documented, and they are simply not on the pricing page — which is the point.

Specs and fine print · 11 min read · Last reviewed 2026-07-29

Key takeaways

"Unmetered" means no per-gigabyte charge. It does not mean unlimited — the limit is enforced elsewhere.
The real constraints on shared hosting are CPU seconds, concurrent processes, memory and I/O throughput.
Inode limits cap the number of files, and are what "unlimited storage" plans actually enforce.
Fair-use clauses let the host suspend any account whose usage they consider disproportionate, at their discretion.
All of this is documented — in the terms of service and knowledge base, not on the pricing page.

Shared hosting plans advertise unlimited bandwidth, unlimited storage, unlimited sites, unlimited databases and unlimited email accounts. None of these are unlimited. All of them are bounded by limits that genuinely exist, are genuinely documented, and are genuinely not on the page you buy from.

This is not fraud; it is a well-established convention that every host follows because the first one to stop would look expensive by comparison. Reading it accurately is a skill.

What "unmetered bandwidth" actually means

It means the host does not bill per gigabyte transferred. That is a real and useful commitment — you will not get a surprise invoice after a traffic spike, which is a genuine risk on metered cloud platforms.

What it does not mean is that you can transfer unlimited data. The constraint simply moves to a different resource, because bandwidth is rarely what actually runs out on a shared server. What runs out is the machine's capacity to generate the responses.

Serving a page requires PHP execution, database queries, memory and disk I/O. A site busy enough to move serious bandwidth is, by definition, consuming a lot of those first. The host limits those, so bandwidth never needs a limit.

The limits that actually bite

CPU seconds and entry processes

The most common real constraint. Shared hosts allocate each account a quota of CPU time and a cap on concurrent PHP processes ("entry processes"). Exceed the process cap and further visitors queue or receive a 508 error while existing requests finish.

This is what people are actually experiencing when a shared-hosted site "goes down under load". The server is fine; your account hit its ceiling.

These figures are usually in the knowledge base rather than the pricing page, and they are the single most useful number to find when comparing shared plans. They correlate with how much simultaneous traffic you can serve far better than storage or bandwidth do.

Inodes

The one that catches people out on "unlimited storage" plans. An inode is roughly one file or directory, and hosts cap the count regardless of total size — commonly in the low hundreds of thousands.

So you can store unlimited gigabytes, provided they are in a limited number of files. A WordPress install with a few plugins is already tens of thousands of inodes. Email is worse: on many hosts each message is a separate file, so a busy mailbox burns inodes quickly. Cache directories are worse still.

Hitting the inode limit typically stops you creating new files — which breaks uploads, backups, sessions and email delivery simultaneously, often presenting as a series of unrelated-looking failures.

Memory and I/O

Per-account memory limits cap what a single PHP process can allocate; memory-hungry plugins hit this and fail with a white screen. I/O throughput limits cap disk read/write speed, and are usually what makes a database-heavy site feel sluggish on shared hosting even when nothing is technically over quota.

Database limits

"Unlimited databases" usually coexists with limits on database size, concurrent connections, or query execution time. For most sites the connection limit matters most, since it caps genuine concurrency.

Fair use, and what it permits

Every unlimited plan is governed by a fair-use or acceptable-use clause. The wording varies; the substance is consistent: the host may throttle, require an upgrade, or suspend any account whose resource consumption they judge disproportionate to normal website use.

Two things follow. It is discretionary — the threshold is whatever the host decides in context. And it is the actual limit on every "unlimited" resource, which is why the specific numbers can be left off the pricing page without anyone being misled in a legally meaningful sense.

In practice hosts apply it reasonably to ordinary websites. It gets invoked against file storage, media libraries used as a CDN for other sites, backup dumping, and sites that have simply outgrown shared hosting. See what hosting is not for.

Where to find the real numbers

Before buying, look for these in this order:

The knowledge base, searching for "resource limits", "inodes", "entry processes" or "CPU limits". This is where hosts document the actual figures, because support needs them.
The terms of service or AUP, for the fair-use wording and what triggers suspension.
The plan comparison table, one tier up. Higher tiers frequently list explicit limits ("2 GB RAM, 200,000 inodes") that the entry tier omits — which tells you the entry tier's are lower.
Pre-sales chat. Ask directly what the inode and entry process limits are on the plan you are considering. If they will not say, that is informative.

Reading between the tiers

A useful trick: compare what a host says about its own plans rather than what it says about the market.

If the entry plan says "unlimited storage" and the tier above says "unlimited storage, higher resource allocation", the difference between them is the limit — they have just declined to quantify it. When both plans claim unlimited everything and differ only in price, the difference is entirely in the undisclosed quotas.

Hosts that publish specific numbers are generally more trustworthy on this, not because they are more generous but because you can plan against a number.

What this means in practice

Do not choose on unlimited claims. They are identical across providers and therefore carry no information. Choose on the numbers that are disclosed — renewal price, inodes, entry processes, memory — and on what the host will tell you when asked.

Watch inodes if you run WordPress or email. Both generate large numbers of small files. Cleaning out old backups, cache and spam is real maintenance on shared hosting, not housekeeping.

Treat entry processes as your concurrency budget. If you expect meaningful simultaneous traffic, this is the number that decides whether shared hosting will hold up.

Know when to leave. Persistent 508 errors, or support suggesting an upgrade under load, mean you have outgrown the tier. That is a normal outcome, not a failure — see choosing a hosting type.

Try it

The comparison table records storage and bandwidth as each provider states them, including where that statement is "unmetered". Read those entries as marketing claims — the useful comparison is renewal price and the drawbacks column, which is where documented limitations appear.