Performance guide · Southampton
Southampton: when the page cannot be cached at all
Sailings, availability, tracking, tide tables. If the content changes every few minutes, most caching advice does not apply to you — and almost nobody writes for that case.
The short answer
When a page must show live data, the cache is not the lever; the query is. Set a time to first byte budget, count the queries behind the template, and cache at a shorter interval rather than not at all — even sixty seconds of micro-caching removes most of the load from a page that changes every few minutes.
HostingFast runs NVMe storage behind LiteSpeed from London, with stated database allowances and per-site PHP versions, so the uncacheable path has real hardware under it rather than a shared spindle and hope.
By the HostingFast team · Reviewed 12 August 2026
24/7
Answers at any hour
1-click
WordPress deploy
Free
TLS, auto-renewing
Daily
Backups, every plan
Ports run to a timetable, and so do the sites that serve them. The freight, marine and travel businesses here publish the state of something that is moving, which is the one kind of page a cache cannot simply hold and hand out again.
That is a genuinely different problem from the one most speed guides address. Everything below is written for pages that have to be assembled on request, quickly, every single time.
Live data pages break the usual advice
A brochure page can be cached for a day and served without loading PHP. A sailings table cannot, because a stale one is worse than a slow one. That leaves you with the request path itself: PHP execution, query count, and whatever external API the data comes from.
Start by deciding how stale is acceptable. Very often the honest answer is thirty or sixty seconds, and micro-caching at that interval turns a thousand requests into one query without anybody noticing. That single decision is usually worth more than every other change combined.
Query cost, connections and the database
Count the queries behind the template before assuming the server. A schedule view that runs one query per row will fall over at exactly the point the schedule gets busy, and that pattern is depressingly common in bespoke plugins. Fix the N+1, add the index, then re-measure.
Allowances are stated rather than implied — 25 MySQL databases on the entry tier, 50 in the middle, 100 at the top of shared, all on NVMe storage. Connection handling matters as much as count: a long-running external API call inside a request holds resources for its whole duration, so give it a timeout you have chosen deliberately.
Set a TTFB budget and hold it
Pick a number for the uncacheable template — say, a figure you can defend to whoever owns the page — and measure against it on every release. Google's field threshold treats 800 ms or less as good, which is a reasonable ceiling for a dynamic page and a poor target for a cached one.
Measure it with curl from a connection you use regularly, three runs, median. The value of a budget is that it turns a vague complaint into a pass or a fail, and a fail into a specific piece of work.
The domain, the mail, and what speed work breaks
.co.uk is what a British freight or survey firm wants on a quotation, and it registers beside the hosting with the records aimed correctly. Mailboxes on your own domain come with the plan, with outbound delivery handled by MailChannels.
The failure to watch for is a DNS change that fixes the website and strands the email. Check the mail records before you touch delegation, not after — a slow page loses a visit, and lost mail loses a contract.
Hardware where caching cannot help
NVMe storage on every tier, stated CPU and memory allocations, per-site PHP version switching and a low accounts-per-server count. When a request has to be built from scratch, those are the things that decide how long it takes.
Daily backups restore from the panel in minutes, and plan changes happen in place with no migration, so a capacity decision does not turn into a project.
- .co.uk registered and renewed alongside the plan
- Stated CPU, RAM and database allowances per tier
- Per-site PHP version switching from the control panel
- Mailboxes on your own domain, with outbound delivery handled
Why HostingFast
Standard on every plan
Real hardware for uncached work
NVMe storage and a restrained accounts-per-server count keep requests that must run PHP and MySQL from queueing behind a neighbour.
Stated allowances
CPU, memory and database counts are published per tier, so a dynamic application can be planned against numbers rather than adjectives.
PHP version per site
Switch a site's version in the control panel and measure the difference, without a legacy application holding everything else back.
Quotations that reach the inbox
Mailboxes are part of the plan and outbound delivery goes through MailChannels, so a schedule confirmation is not sitting in a junk folder while a vessel waits on it.
In-place upgrades
Plan changes run from the client area with no migration and no downtime, from shared through VPS to dedicated.
Support at any hour
Engineers on shift around the clock, first reply inside two hours and inside one when a site is down. Shipping does not keep office hours.
Quick Start
From order to online
- 1
Decide how stale is acceptable
Thirty seconds, sixty, five minutes. Whatever it is, micro-caching at that interval is the single biggest win available to a live page.
- 2
Count the queries
Profile the template. One query per row is the pattern to look for, and it is the reason the page dies exactly when it gets busy.
- 3
Time the external calls
Any API inside the request holds resources for its whole duration. Give it a deliberate timeout and a fallback.
- 4
Set the budget and test on release
A TTFB figure for the template, measured three times with the median kept, checked every time the page changes.
Built In
Loaded onto every plan
- An agreed staleness window, with micro-caching set to match it
- Query count profiled per template, with N+1 patterns removed
- Deliberate timeouts on every external API call in the request path
- A written TTFB budget for uncacheable templates
- Stated CPU, RAM and database allowances on the plan
- Mail records verified before any DNS change
- Per-site PHP version switching from the control panel
- A daily backup you can restore when a feed change breaks a live template
- In-place account upgrades — no migration when you change plan
- Human support on duty every hour of every day
Frequently Asked
What people ask us most often
Our schedule page changes constantly. Can it be cached at all?
Almost always yes, just briefly. Decide what staleness is acceptable — thirty or sixty seconds is usually fine for a sailings or availability table — and cache for exactly that. It converts a thousand requests into one query, and no visitor can tell the difference between data that is current and data that is fifty seconds old.
How do I know whether the database or the API is the problem?
Instrument both. Log the duration of the external call separately from the query time and compare them over a day. If the API dominates, the fix is a timeout, a fallback and a short cache in front of it. If the queries dominate, look for one query per row and the missing index behind it.
What database allowance do the plans carry?
25 MySQL databases on the entry tier, 50 in the middle and 100 on the top shared plan, all on NVMe storage. If a dynamic application needs more headroom than that, the account upgrades in place with no migration and no downtime.
What is the risk to our email when we move hosts?
That the mail records get overlooked in the excitement of the website working. Check MX, SPF and any DKIM records before you change delegation, not after, and keep the old mailboxes reachable until you have confirmed delivery both ways. A slow page costs a visit; stranded mail costs a contract.
Keep reading
Web Hosting in Karachi
Pakistani connections, and what 120-200 ms means for a page that must still feel quick.
Web Hosting in Cambridge
Deep-tech spin-outs, documentation sites and an audience that is often not in Britain.
Domain and Hosting on One Order
Buying a name and hosting together, and what that saves you in DNS work.
Web Hosting
cPanel hosting on NVMe behind LiteSpeed, with SSL, migration and a year-one domain.
Reseller Hosting
Reseller accounts with WHM, client isolation and your own branding on top.
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.
Fast, even when nothing can be cached.
NVMe storage, stated allocations and per-site PHP versions — with mail on your own domain and support at any hour.
View Web Hosting plans