21+ years in hosting · Jaipur, India · INR billing with 18% GST · 24/7 email support
Blog · WebJi® India

How to Prevent a Website Crash From Traffic: Triage Order

Quick answer

A traffic crash happens when requests exhaust CPU, RAM, database connections or PHP workers faster than the server can answer. Read the error code first, check CPU and database connections in the panel, then add page cache and a CDN before you buy a bigger plan.

For site owners whose website slows down or errors during campaigns, sales or viral moments, and who want to fix the cause before paying for more.

✓ In business since 2005
✓ INR + GST billing
✓ 24/7 Email Support — avg reply < 60 min
GuideWebJi®
knowledge
Read time
4 min
Sections
8
Updated
14 Nov 2026
Written by
WebJi® team

Why does a traffic spike crash a website?

Each visit asks the server for PHP work, a database row and a few files. A campaign multiplies those asks within a minute. When CPU, RAM, database connections or PHP workers run out, the next visitor waits and then gets an error. The site did not lose its files; it ran out of room to answer.

  • Uncached PHP running on every hit.
  • A plugin that queries the database on the homepage.
  • Images large enough that the network port fills before CPU does.
Indian trader monitoring candlestick charts on multiple screens in evening light

What should you check first when traffic is crashing the site?

Read the error the visitor sees before you touch a setting. A 503 or 508 means the worker or entry-process limit is full, a 500 means PHP hit a fatal error (see fixing a 503 error for the resource-limit case), and “Error establishing a database connection” means MySQL has no free connections.

  1. Load the site from another network to confirm the spike is real and not one bad connection.
  2. Note the exact error code and the page it appears on.
  3. Open the host panel and read CPU, RAM, entry processes and MySQL connections together.
  4. Turn on page cache for public pages and pause any plugin added in the last day.
  5. Check access logs for one IP or one crawler hitting hard; a scraper looks like traffic.
  6. Reload after each single change, and write down which change moved the number.

Only after those steps should you price a bigger machine. Changing plan mid-spike usually moves the site to a new server with the same slow query and the same missing cache. WebJi® publishes a 99.9% uptime target; that is a target for the platform, and a site that exhausts its own CPU limit still returns errors.

How does caching stop repeat visits from crashing the site?

A page cache stores the finished HTML, so the second visitor gets that copy instead of another PHP run. Carts, checkout and logged-in pages stay uncached, otherwise a person can see someone else’s basket. An object cache such as Redis takes repeated database reads out of the loop, and PHP OPcache removes the cost of recompiling PHP files.

A CDN keeps images and cached HTML closer to the visitor, which is why the origin sees fewer repeats. Watch the cache hit ratio for a week: a low ratio means the cache is being bypassed, and no plan size fixes that.

Why fix images and slow queries before adding servers?

Upload images at the size they are displayed, then compress and lazy-load them. Full-size photos are the usual reason a site that handled 200 visitors fails at 2,000, because the network port fills while CPU is still idle.

On the database side, read the slow query log for the week of the spike and index the columns you filter on. A page that fires one query per product row, the N+1 pattern, will use any extra CPU you buy, so fix the query loop before the plan.

How to prepare a website for a sale or campaign spike

Warm the cache the day before, run a load test above the traffic you expect, and put the checkout through that test rather than only the homepage. If a normal Tuesday already sits near the CPU cap, move plan before the sale, not during it.

Write the campaign date, expected peak visitors per minute and the current cap in one line, and send it to whoever runs the ads. If the ads go out at 9 am, the plan change has to be finished the night before. The escalation path for the day itself is in what to do when the website is down.

When is shared hosting the wrong size for your traffic?

Shared hosting is the wrong size when a normal day already sits near the CPU or entry-process cap, or when a single campaign can double your peak. In WebJi®’s published pricing, the shared plans Webji Launch, Turbo and Scale are ₹299, ₹399 and ₹699 a month, and Cloud Entry (2 vCPU, 8 GB RAM) is ₹5,900 a month, all before 18% GST; VPS configurations are quoted, and checkout confirms the final amount.

A VPS gives a reserved slice with root access, and VPS hosting lists the sizes. Cloud hosting is the step after that, for a machine you can resize without a migration. Both are quoted with a date and GST on the invoice.

What should you change after a traffic crash?

Write down four numbers from the incident: peak concurrent visitors, error rate at the peak, the slowest query, and the cache hit ratio. Those four decide the next plan size, and they also show whether the fix worked when the next campaign arrives.

Then change one thing at a time: cache rule, index, image sizes, plan. A post-incident note of half a page is enough, as long as it names the cause rather than the symptom.

Capacity ranges for each tier, and a spike-day playbook, are in the best hosting for high-traffic sites.

Sources: CloudLinux documentation for 508 “Resource Limit Is Reached” and entry processes; MySQL reference manual for the slow query log; WebJi® published pricing.

FAQ

Frequently asked questions

What causes a website to crash from traffic?

Requests use up CPU, RAM, database connections or PHP workers faster than the server can answer. Uncached pages, database queries on every load and full-size images make that happen sooner.

What should I check first during a spike?

Read the error code: 503 or 508 means the worker or entry-process limit is full, 500 means a PHP fatal error, and a database connection error means MySQL connections are gone. Then read CPU, RAM and connections in the panel before changing anything.

Does a CDN stop a crash?

A CDN serves repeat reads of images and cached HTML, so the origin answers fewer times and survives a bigger spike. Checkout, login and cart pages still reach the origin, so they remain the limit.

When should I move off shared hosting?

Move when a normal day already sits near the CPU or entry-process cap, or when one campaign can double your peak. In WebJi®’s published pricing, the shared plans Webji Launch, Turbo and Scale are ₹299, ₹399 and ₹699 a month, and Cloud Entry (2 vCPU, 8 GB RAM) is ₹5,900 a month, all before 18% GST; VPS configurations are quoted, and checkout confirms the final amount.

Will a bigger plan fix a slow database?

Not by itself. A slow query or a missing index will use the extra CPU too, so read the slow query log and add the index first. A bigger plan buys headroom, it does not rewrite a bad query.

Last editorial update: 15 November 2026. Prices follow the dated sources on this page; checkout confirms current charges.

WebJi® since 2005

Ready to get started with WebJi®?

Join 12,000+ businesses across India and overseas. INR pricing, 18% GST shown before payment and 24/7 email support in Hindi and English — average first reply under 60 minutes.