Hosting resource limits: what to check before you buy

Unlimited disk still sells. The thing that knocks a site over is CPU seconds, inodes, and PHP workers the pricing table never mentions. Before you buy, get the real limits in writing.

I have had a "perfectly fine" shared plan fall over on a quiet Tuesday because a backup, a plugin update, and a handful of cron jobs all tried to run at once. The homepage still loaded. Checkout and wp-admin sat there spinning. Support said the site was using too many resources. They would not say how many I was allowed. If you are already comparing hosts without leaning on top 10 roundups, pair this with what overselling meant in 2012 and how to choose web hosting without relying on top 10 lists. For why sales copy and review copy so often rhyme, read The Truth About Hosting Review Sites.

Unlimited is a brochure word

Disk and bandwidth used to be the numbers on the box. Hosts still print them in huge type. They are rarely what actually throttles you.

The real caps sit one layer down:

CPU seconds or a CPU percentage per account
inode count (how many files you can have, not how many gigabytes)
PHP workers / entry processes (how many requests can run at the same time)
I/O wait (how hard the disk will let you hit it)
process count, including cron and SSH

I wrote about overselling years ago as packing too many clients onto one box. That is still the business model. The packaging changed. "Unlimited" now means they will not quote a disk number, then they will throttle CPU when the neighbour's site has a sale. Unlimited is whatever support will not raise without a paid move.

A plan that brags about unlimited sites, unlimited databases, and unlimited mailboxes on a machine that will not name its CPU cap is selling a neighbourhood, not a promise. The shared vs VPS post is what you read after you have hit that wall. This one is how you see the wall before you pay.

Ask for numbers, not adjectives

Before you buy, send the same short question to every host on your shortlist:

On this plan, what is the CPU limit, inode limit, PHP memory limit, and how many PHP workers or entry processes do I get? What happens when I hit them: throttle, 508, 503, or a suspension? Can I see usage in the panel, or only after a ticket? If I need a second copy of the site for staging, is there headroom for the files and the database?

Save the reply. "Resources are generous" is not a number. "We rarely have issues" is not a number either.

Push on the boring edges:

Is the CPU limit a fair-share percentage, a hard second count per day, or CloudLinux LVE numbers you can actually see?
Do inodes include mail, backups they keep on the account, and cache files, or only "your" files?
Are PHP workers dedicated, or a pool you share with every other account on the box?
Does a backup they take count against your I/O and CPU while it runs?
If you host WordPress or WooCommerce, do they quote workers for admin and checkout, or only for "normal pages"?

A host that will paste the LVE / cPanel resource page for that plan is taking you seriously. A host that says "just try it, you can always upgrade" is counting on you to pay twice.

How you find out you already hit them

You do not need a traffic spike that looks like a product launch. The cheap tells are smaller:

The public site is fine and wp-admin or checkout is slow
Cron jobs skip, backups never finish, or staging cannot clone because inodes are gone
Error logs mention 508 resource limit, 503, or "resource temporarily unavailable"
Support blames plugins every time, but will not name the worker or CPU cap on the plan you bought
A quiet hour still trips the limit because a backup and an indexer ran together

If images are huge or a page builder is doing PHP on every request, more CPU will not save you. It will just be a more expensive place to be slow. Fix the site first, or at least know which problem you are solving.

A VPS is not a magic wand

When the shared cap is real and the site is actually busy, a VPS can help: reserved CPU and RAM, your own PHP-FPM workers, room for Redis or a queue. It also hands you the adult jobs: updates, firewalls, watching disk, not filling the box with a second copy you forgot about.

A tiny VPS with bad disk and no cache can feel worse than a decent shared plan. So can a "managed VPS" that still hides the same inode and backup limits behind a friendlier panel. Ask for the numbers there too. The upgrade is only useful if you can see what you bought.

Red flags worth pausing on

Be careful when:

Unlimited is on the pricing table and CPU, inodes, and workers are "available on request"
Chat will talk about disk and bandwidth all day and go quiet when you ask about LVE or entry processes
The panel has no resource graph until after they suspend you
Backups they keep on the account eat the inode budget, then they tell you to upgrade
Reviews praise "unlimited" but mention 508s, surprise throttling, or a move they were pushed into after a busy weekend

Useful hosts treat limits as part of the product: a number you can see, a throttle you can explain, and a path that is not "buy the next tier because we packed the last one." Useful reviews mention the Tuesday afternoon, not only the signup bonus. If neither does, you are reading advertising.

We care about this for the same reason we care about overselling honesty and uptime SLAs: the real product is a site that stays responsive when two jobs run at once, not a badge that says unlimited. Our FAQ on how we rank hosts prefers evidence over splash ratings, and a resource story you cannot get in writing is just another sticker.

The best course of action is still the boring one. Ask the same resource questions of every shortlisted host, try a restore and a busy-hour admin click on the trial if they offer one, and treat any unlimited pitch as marketing until they show you the graph. Browse more web hosting guides and keep the homepage ranking list handy while you compare.