Handling holiday peaks for a seasonal retailer

A shop's website slows down in its busiest weeks. Smarter caching and servers that grow with demand handle the rush without paying for holiday capacity all year.

2 min readMade-up stories, not client work

Scalability: story 10 of 15

A made-up story

Picture an online gift shop that makes much of its yearly revenue in a few weeks before the holidays.

What's going wrong

Every year, the site slows to a crawl just when shoppers arrive. The usual fix is renting bigger servers for the whole year.

What they have today

  • A store on cloud servers
  • Product photos served from the same servers
  • A database that gets busy at peak times

What we can't ignore

  • The busiest weeks can't be risked
  • The budget is planned around the season
  • A small team

The choices

  • Worth a look

    Rent bigger servers for the season

    Simple, but it pays for peak size even on quiet days.

  • Our pick

    Cache pages and photos closer to shoppers

    Most visitors see the same pages, so they can be served without touching the servers.

  • Our pick

    Servers that add copies when busy

    Extra copies appear during the rush and go away afterwards.

  • Not for them

    Rebuild the store from scratch

    Far too much risk before the biggest season.

What we'd do

Put a caching service in front of the store, and move photos to cheap storage behind it. Let the servers add copies when busy and remove them when quiet. Test with fake shoppers well before the season, and add a second database only if the test shows it's needed.

Why

  • Most of the load disappears with caching
  • Extra capacity only when it's needed
  • Testing before the season avoids surprises

Where the money really goes

No prices here: this shows where the costs sit, not what they'd be for any real business.

Caching is inexpensive and cuts the load on everything behind it. Extra servers only cost money during the rush.

Caching
A modest monthly fee.
Servers
A small base, plus extra copies only during peaks.
Lost sales
A slow site at peak time costs more than any of the above.

Running, growing, and staying safe

If something breaks

If one server copy fails, others carry on, and cached pages keep showing products.

When it grows

More shoppers mean more copies, automatically, up to a set limit.

Keeping it safe

The caching service also absorbs junk traffic before it reaches the store.

What the business gets

  • A fast site during the biggest weeks
  • No paying for peak size all year
  • A tested plan before the rush

These are the kinds of results the approach aims for, not measured results.

When we'd look again

If growth continues and the database is still the bottleneck, we'd look at caching more of the catalog or reworking the slowest lookups.

If you liked this, read

About these stories. These stories are made up to show how we think. They are not client projects, describe no real business, and contain no measured results.

Next: Ticket sale spikes for an events promoter

Huge crowds for a few minutes, then quiet for weeks. A virtual waiting line protects the site far more cheaply than building for the first five minutes of every sale.

Read the next story