Most Of A Spike Never Reaches The Server.
Cloudflare answers the cacheable pages at the edge, the Nginx page cache answers what gets past it, and Redis keeps the same queries off MySQL. What is left is the traffic that genuinely has to run, in a container that is yours alone.
Why Busy Sites Fall Over
Three things, and only one of them is about how many visitors you had.
A Spike Is Usually One Page
A newsletter, a post that travels or a campaign sends thousands of people to the same handful of URLs. Served from cache that is close to free. Rebuilt by PHP for every visitor it is the thing that takes the site down.
The Database Gives Out First
On a busy WordPress site the limit is rarely bandwidth. It is PHP workers waiting on the same queries over and over, until requests queue up behind each other and the whole site feels slow.
Neighbours Make It Worse
On shared hosting your busy afternoon competes with everyone else’s, and theirs competes with yours. You cannot plan around capacity you cannot see.
Four Layers, In This Order
The edge, the page cache, the object cache and then your own container. Every bit of it is on every plan, with nothing to buy and nothing to set up.
Cloudflare Takes The First Hit
Cloudflare sits in front of every site, serving static files and cached pages from the edge close to each visitor and absorbing DDoS traffic. For a spike on cacheable pages, most of the requests are answered before anything reaches the server at all.
An Nginx Page Cache Behind It
What does reach the origin is served from the Nginx page cache rather than being rebuilt by WordPress. One visitor pays for the page, the rest get a copy. There is no caching plugin to configure and nothing to tune.
Redis Keeps MySQL Out Of It
The object cache holds the lookups WordPress repeats on every request in memory, so MySQL 8.4 is asked once instead of thousands of times. This is the single biggest difference on a site with a lot of logged-in traffic.
Your Spike Is Yours
Every site runs in its own container, as its own Unix user, with its own MySQL database. Your busiest hour is not sharing a pool with a stranger’s busiest hour, and your own sites do not fight each other either.
Current PHP, Per Site
PHP 8.3 or 8.4 chosen per site. On the requests that genuinely have to run, the version doing the running matters, and you move when you are ready rather than when a shared pool is upgraded underneath you.
Cron That Does Not Need Visitors
WordPress cron runs from a real system cron instead of firing on page loads. Scheduled jobs do not pile onto a visitor’s request during your busiest hour, and they still run when the site is quiet.
You Can See What You Are Using
Storage and bandwidth per site in the members area, with per-site caps available inside your pooled account allowance. Real numbers, so a busy month is something you watch rather than something you discover.
One Machine Only Goes So Far
Caching moves the ceiling a long way out, and for most sites that is the end of the story. It does not remove the ceiling. The requests that cannot be cached still land on one machine in Singapore, shared with other customers’ containers, and eventually that is the limit you meet.
At that point tuning stops helping and the answer is a machine of your own: a Google Cloud instance in Singapore running nothing but your sites, with its own database instance. Enrollment for dedicated machines is temporarily closed, so if you are heading that way it is worth telling us early rather than finding out in the middle of a campaign.
There is no automatic failover between machines here and no published uptime SLA. Recovery from a lost machine is a rebuild from the nightly backup, which is kept in Google Cloud Storage away from the machines. Worth knowing before a launch, not after.
About Dedicated Machines
Scale Has The Most Room
Ten websites, 30 GB of storage and 200 GB of bandwidth a month, pooled across the account. USD 80 a month billed yearly, or 96 billed monthly.
Scale
$960 a year. That is $192 less than paying monthly.
- Up to 10 websites
- 30 GB storage
- 200 GB bandwidth a month
- A temporary address to test on
The Most Headroom On A Shared Plan
Up to ten websites sharing one pool of resources. For agencies and studios with a client list.
- Servers In Singapore
- Free SSL On Every Domain
- Daily Backups, Kept 30 Days
- One-Click Staging
- Cloudflare CDN And DDoS Protection
- Nightly Malware Scans
- SFTP And WP-CLI
- PHP 8.3 And Redis
- Support From Real People
Every plan includes the same platform. The only difference is how many websites and how much room they have.
A single busy site does not need ten slots. The help centre explains what counts as bandwidth and how the caching works before you pick.
What Busy Sites Ask First
How Much Traffic Can A Site Here Take?
It depends almost entirely on how much of your traffic can be cached. A news site or a blog where nearly every request is an anonymous visitor reading a page is carried mostly by Cloudflare and the page cache, and the server barely notices. A shop or a membership site, where a large share of requests are logged in or in a checkout, is limited by PHP and the database instead. We will not print a visitors-per-month figure, because the honest answer is a different number for every site.
Which Pages Cannot Be Cached?
Carts, checkouts, account pages, anything served to a logged-in session and internal search. Those go to PHP and the database on every request, which is also why a burst of search traffic or a flash sale is much heavier than the same number of people reading an article.
Is The Bandwidth Unlimited?
No. It is 50 GB a month on Build, 100 on Grow and 200 on Scale, pooled across the account, and your usage is visible per site. Nobody selling unlimited bandwidth means it, and we would rather give you a number you can plan against.
Does It Scale Itself When Traffic Arrives?
No. There is no autoscaling here, and no bill that grows while you sleep. A plan is a plan. The upside is that an unexpected spike cannot cost you a fortune on a Sunday, and the downside is that there is a ceiling, which the next answer is about.
What Happens When A Site Outgrows The Machine?
A single machine has a ceiling, and once the uncacheable share of your traffic gets close to it the right answer stops being tuning and becomes a machine of your own. That is what dedicated hosting is for, and enrollment for it is temporarily closed while we finish the groundwork. If you are near that point, talk to us before you buy a plan.
Is There A High Availability Setup?
No, and it would be wrong to imply one. There is no automatic failover between machines: if a machine is lost, sites are rebuilt from the nightly backup held in Google Cloud Storage. There is no published uptime SLA either. If your traffic comes with an availability guarantee attached, that is a conversation to have with us rather than an assumption to make.
What We Do Not Do
Not Our Job, And Not Promised
- We do not autoscale, and the bill does not scale itself. A plan has a fixed price and a fixed allowance, and a machine has a ceiling.
- We do not sell unlimited bandwidth. Build is 50 GB a month, Grow 100 and Scale 200, pooled across the account and visible per site.
- We do not cache the uncacheable. Carts, checkouts, logged-in sessions and internal search go to PHP and the database every time, and no host changes that.
- We do not fail over to another machine, and we publish no uptime SLA. Recovery from a lost machine is a rebuild from the nightly backup.
- We do not rewrite your theme or your plugins. We will tell you what the server saw, but a page that loads forty scripts stays a page that loads forty scripts.
- We do not move busy sites across for you. Migration is self-service, onto a temporary address you can test on first.
Try It With Your Own Traffic
Bring a copy of the site onto a temporary address, point a share of your traffic at it, and see what the cache does before you move the domain.
Start On Scale