Performance guide · Oxford
Oxford: thousands of URLs, one cache to fill
A site with forty pages behaves nothing like a site with forty thousand. Long-tail archives break the assumptions behind most caching advice, and nobody warns you.
The short answer
On a large archive the problem is not making one page fast — it is that most pages are requested rarely, so they are almost never in the cache when somebody asks. That makes the cold, uncached response time the number that actually describes your site.
HostingFast runs NVMe storage behind LiteSpeed from London, with stated storage allowances up to 200 GB on the top shared tier and PHP X-Ray on CloudLinux Pro to profile the cold path directly.
By the HostingFast team · Reviewed 12 August 2026
99.9%
Uptime target, monitored
24/7
Engineers on shift
Free
SSL, renewed for you
NVMe
Storage under the archive
The sites here are archives before they are websites: two decades of articles, a catalogue of titles, conference papers and course pages that nobody has deleted, because nobody should.
Publishing at that scale produces sites with very long tails: thousands of pages, most of them visited occasionally. Everything below follows from that single structural fact.
Long tails break the usual caching assumption
Cache advice is written for sites where the same twenty pages get most of the traffic. On an archive of thousands, a given page may be requested once a week, which means it is almost certainly cold when a reader arrives. The cache still helps your popular pages; it does very little for the tail.
So measure the cold path deliberately. Purge, request a page nobody has opened recently, and read the time to first byte on that request. That number is what most of your readers actually experience, and it is the one nobody tests.
Static content is the easiest win there is
The upside of an archive is that the content rarely changes, which makes long cache lifetimes safe and makes pre-warming worthwhile. A crawl of your own sitemap after a deployment fills the cache before readers do, and it costs nothing but a few minutes of server time.
Set long expiry on assets, keep the CSS and font payload small since it is amortised across thousands of pages, and avoid per-page scripts. On a large site, a 30 KB saving on the shared bundle is worth more than any single page optimisation.
Where the milliseconds hide on a publishing site
Three places, in order of frequency: a related-articles or archive widget running an expensive query on every page, a search feature scanning content columns without an index, and an image pipeline resizing on request rather than in advance. Each is invisible on a small site and decisive on a large one.
PHP X-Ray on CloudLinux Pro at the top shared tier profiles a live request and names the call, which is considerably faster than disabling widgets one at a time across a site with forty thousand pages.
When storage and capacity genuinely need a bigger tier
Sometimes the answer really is more room. Storage runs 10 GB on the entry plan, 20 GB in the middle and 200 GB on the top shared tier, with backups every six hours rather than daily at the top and Imunify360 in front. An archive with decades of PDFs will find the ceiling honestly.
Plan changes happen in place from the client area with no migration and no downtime, so the upgrade is a decision rather than a project — and it should still be made on the evidence of a profile rather than a hunch.
Suited to sites that keep growing
NVMe storage with LiteSpeed caching in the web server, 200 GB available on the top shared tier, and profiling on the same plan so a cold-path problem can be diagnosed rather than guessed at.
Backups land every six hours on the top tier with Imunify360 in front, and plan changes apply in place with no migration when the archive outgrows its allowance.
- .co.uk or .com registered and renewed with the plan
- Up to 200 GB of NVMe storage on the top shared tier
- Backups every six hours on the top tier
- In-place upgrades, no migration when you change plan
Why HostingFast
Standard on every plan
Cold-path performance
NVMe storage and a low accounts-per-server count decide how fast an uncached page is built, which on a long-tail archive is most of them.
Profiling on the same account
PHP X-Ray on CloudLinux Pro at the top shared tier names the expensive call instead of leaving you to bisect widgets across thousands of pages.
Storage that scales
10 GB, 20 GB or 200 GB of NVMe depending on tier, so an archive of documents has somewhere honest to grow into.
More frequent backups at the top
The top shared tier backs up every six hours rather than daily, with Imunify360 in front of it.
Upgrade without migrating
Plan changes run from the client area with no migration and no downtime, which matters when the site is large enough to make moving painful.
Uptime with a credit attached
A 99.9% target under constant monitoring, with a pro-rated credit if a month falls short.
Quick Start
From order to online
- 1
Measure the cold path
Purge, request a rarely-visited page, and record the time to first byte. That is the number most of your readers meet.
- 2
Shrink the shared bundle
CSS, fonts and scripts are paid on every page. On a large site, trimming the shared payload beats optimising any single page.
- 3
Profile the expensive widget
Related articles, archive lists and site search are the usual suspects. Find the query rather than removing features at random.
- 4
Pre-warm after deploys
Crawl your own sitemap once a release is live so the cache is populated before readers arrive rather than by them.
Built In
Loaded onto every plan
- Cold, uncached time to first byte measured deliberately
- Shared CSS, font and script payload kept small and audited
- Related-articles and archive queries profiled for cost
- Site search running against an index, not a content scan
- Cache pre-warmed by crawling the sitemap after a deploy
- Stated storage allowance checked against the archive's growth
- A daily backup of an archive nobody wants to rebuild from memory
- In-place account upgrades — no migration when you change plan
- NVMe SSD storage on every tier, not just the expensive ones
- 99.9% uptime as the target, monitored around the clock
Frequently Asked
What people ask us most often
Why does our archive feel slow when the home page is quick?
Because the home page is always in the cache and the archive almost never is. A page requested once a week is cold when a reader arrives, so it pays for the full PHP and database path. Measure that cold response deliberately — purge, request an obscure page, read the TTFB — and optimise against that number instead.
Is pre-warming the cache worth the effort?
On a large, rarely-changing site, yes. Crawling your own sitemap after a deployment fills the cache in minutes and means the first reader of each page is not the one paying to build it. It is a few lines of scripting and it targets exactly the weakness a long tail creates.
How much storage do the plans give us?
10 GB on the entry tier, 20 GB in the middle and 200 GB on the top shared plan, all NVMe. The top tier also backs up every six hours rather than daily and runs Imunify360 in front. If an archive outgrows that, the account upgrades in place with no migration and no downtime.
What usually makes a publishing site slow on the cold path?
An expensive query rendered on every page — related articles, archive counts, a tag cloud — or site search scanning content without an index. Both are unnoticeable at forty pages and decisive at forty thousand. Profile the request rather than guessing: on the top shared tier, PHP X-Ray on CloudLinux Pro attributes the time to a specific call.
Keep reading
Web Hosting in Southampton
Port and freight sites, live data pages and micro-caching that works.
Web Hosting in Sheffield
Catalogue sites, slow queries and the index that fixes them.
National Endings (ccTLD)
Country-code endings, the rules behind them, and where they help an archive.
Joomla Hosting
Joomla on NVMe with server-level caching and the CLI access you expect.
Domain Names
Find, register and transfer domains — first year free with annual hosting.
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 on the pages nobody tests.
NVMe under a server-level cache, up to 200 GB of storage, and profiling on the same plan so the cold path can be fixed.
View Joomla Hosting plans