Stack Glossary
Shared hosting, judged by what the neighbours can do to you
The plan looks identical to every other shared plan on paper, and yet one of them is quick at 4pm and the other is not.
The short answer
Shared hosting runs many accounts on one physical machine, each fenced inside its own resource limits — CPU, memory, processes and I/O — with the cost of the hardware and the network divided across everyone on board. What decides whether it feels quick is the density policy, not the word shared.
Two providers can advertise the same specification and deliver very different results, because one puts a sane number of accounts on a box and the other does not. That difference shows up as variance across the day rather than as a bad average, which is exactly why a single speed test tells you almost nothing.
Every shared tier here sits on NVMe storage behind a LiteSpeed cache, from the entry Sprint tier to the largest Nitro one.
By the HostingFast team · Reviewed 12 August 2026
100+
Terms defined here
2 min
Typical read time
Plain
English, no hand-waving
24/7
Engineers on shift
Isolation on a modern shared platform is done with kernel-level limits of the CloudLinux sort: each account gets a ceiling on CPU, memory and concurrent processes, and hitting your own ceiling throttles you rather than the machine. That is what stops one account's runaway import from taking the server down with it.
The limits also explain the two error patterns people meet. A site that stalls under its own traffic is usually hitting its entry-process or memory ceiling, and the fix is caching or a bigger tier. A site that is fine at breakfast and sluggish by mid-afternoon is measuring the machine's total load, which is the host's problem, not yours.
How to test for a noisy neighbour
Pick one URL that you know is cached and time it hourly for a working day. A well-run platform gives you a tight cluster of readings. Contention shows up as a wide spread: quick overnight, slow through the afternoon peak, with the same page and the same code in both cases.
Do the same test against a site the provider already hosts before you buy. It takes three minutes with curl and tells you more than any feature grid, because it measures the thing nobody publishes: how many accounts they are prepared to stack on one machine.
The limits that actually bind
On a shared account the ceiling you meet first is rarely disk. It is concurrent PHP processes and memory, and both are consumed by uncached requests. Ten visitors on a cached page cost the machine almost nothing; ten logged-in visitors hitting an uncacheable dashboard cost it ten PHP processes and every query behind them.
This is why server-level caching matters more on shared hosting than anywhere else. LiteSpeed answering from memory keeps your process count low, which keeps you well inside your own limits during exactly the hour you would otherwise breach them.
When the model genuinely runs out
Move on when you need something the model cannot give: root, a daemon you install yourself, a non-PHP runtime with its own process manager, or resources guaranteed rather than shared. Those are architectural requirements, and no amount of tuning gets around them.
Do not move because a page is slow. Time it first. Slow uncached pages on shared hosting are usually an unindexed query, a plugin calling an external API on every load, or images that were never resized — all of which will follow you onto a more expensive machine unchanged.
How the shared tiers are built here
Sprint carries 10 GB of NVMe, 500 GB of bandwidth, 1 core and 2 GB, and 25 MySQL databases for a single site. Turbo takes it to 25 sites, 20 GB and 1 TB. Nitro runs 100 sites on 200 GB with 2 cores and 4 GB, backups every six hours and PHP X-Ray on CloudLinux Pro for finding slow code.
Read VPS next if you want the guarantee rather than the share, and Cache if you would rather make the shared account behave like a bigger one.

Written for people who read response headers
The interesting question about shared hosting is never the definition. It is how many accounts sit on the machine, what the per-account limits are, and how the platform behaves at 4pm on a Tuesday. Those are the questions this entry answers.
PHP X-Ray on the Nitro tier traces a slow request down to the function, which turns an argument about hosting into a line number.
- Density explained, not glossed over
- The hourly test for contention
- Which limit you hit first, and why
- Same NVMe and LiteSpeed on every tier
Why HostingFast
Standard on every plan
Kernel-level account isolation
CloudLinux-style limits cap CPU, memory and processes per account, so a neighbour's bad afternoon stays their problem.
LiteSpeed on the shared tiers
Cached HTML is served from memory, which keeps your PHP process count low during the busiest hour of the day.
Backups every six hours on Nitro
The top shared tier snapshots four times a day with Imunify360 in front of it, restorable from the panel.
PHP X-Ray on CloudLinux Pro
Available on Nitro, it traces a slow request to the function responsible instead of leaving you to guess.
Per-site PHP releases
One account can run a legacy application on an older 8.x branch and a modern one on the current release.
Upgrades in place
Moving from Sprint to Turbo or Nitro is a plan change on the same account — no migration, no new IP, no downtime.
Quick Start
From order to online
- 1
Time a cached URL through a full day
Hourly readings on one page. A tight cluster is a healthy platform; a wide spread through the afternoon is contention you are paying for.
- 2
Pick the tier by limits, not by disk
Sprint carries one site on 1 core and 2 GB; Nitro carries 100 sites on 2 cores and 4 GB. Sites and headroom decide it, rarely storage.
- 3
Verify the cache is doing its job
After the move, request a public page twice and check for x-litespeed-cache: hit. If it never hits, the win you paid for is not being delivered.
Built In
Loaded onto every plan
- NVMe SSD storage on every shared tier, Sprint included
- LiteSpeed caching built into the server, not bolted on by plugin
- Per-account CPU, memory and process limits
- Per-site PHP 8.x version switching from the panel
- Daily backups, and every six hours on Nitro
- Free SSL, issued and renewed automatically
- Free migration, performed by our engineers, from your current host
- In-place upgrades between Sprint, Turbo and Nitro
- 30-day money-back window on shared plans
- Renewal at the price you signed up at
Frequently Asked
What people ask us most often
How can I tell how many accounts are on my server?
You cannot see the count directly, but you can measure its effect. Time one cached URL every hour for a day. A platform with sane density produces readings that barely move; an oversold one produces a curve that tracks the working day. The second pattern is the only evidence you need.
Does a neighbour getting hacked put my account at risk?
Kernel-level isolation keeps accounts out of one another's filesystems and processes, and server-side malware scanning runs across the machine, so a compromise next door is contained rather than shared. The engineering does that work — not the number of accounts on the box.
Which resource limit do sites usually hit first?
Concurrent entry processes and memory, almost always, and almost always on uncached requests. Disk and bandwidth allowances are rarely the binding constraint on a normal site. Cache the pages that can be cached and the same plan carries several times the traffic.
Is shared hosting fast enough for WooCommerce?
For a modest catalogue, yes, with a caveat worth knowing: cart, checkout and account pages are uncacheable by design, so they run PHP and MySQL every time. Judge a store plan on how it handles those pages under concurrency, not on how quickly the cached home page loads.
Keep reading
cPanel
The panel each shared account is administered through, and the tools on it worth learning first.
Web Hosting
The layer underneath every plan: what serves the request, and what decides how long it takes.
Best Shared vs VPS Hosting
Shared or VPS, decided on measurements rather than instinct — including the test that settles it.
Domain Names
Register or transfer a name, with the first year included on an annual hosting plan.
WordPress Hosting
Managed WordPress with LiteSpeed and Redis in front, plus staging copies and daily backups.
Changing hosts? Run through our checklist first.
A straightforward sequence for a switch your visitors never feel: which files move first, how to shift email across without losing a single message, the right moment to repoint DNS, and the two mistakes behind almost all the downtime we get asked to rescue.
Shared, without the afternoon slowdown.
NVMe and LiteSpeed on every tier, with per-account limits that keep the neighbours out of your response times.
View Domain Names plans