Why should you care?

Flipkart-scale ShopNest runs twelve API pods — in-memory cache is per-node; Redis shares hot product data cluster-wide.

See it live — copy this example

Paste into a .NET project with EF Core packages, then run with LocalDB/SQL Server (dotnet ef / dotnet run).

var key = $"shopnest:product:{id}";
var json = await _redis.GetStringAsync(key);
if (json is not null)
    return JsonSerializer.Deserialize<ProductCardDto>(json);

var dto = await _context.Products.AsNoTracking()
    .Where(p => p.Id == id)
    .Select(p => new ProductCardDto(p.Id, p.Name, p.Price))
    .FirstAsync();

await _redis.SetStringAsync(key, JsonSerializer.Serialize(dto),
    new DistributedCacheEntryOptions { AbsoluteExpirationRelativeToNow = TimeSpan.FromMinutes(10) });
return dto;

What happened?

• Redis checked first.
• On miss EF projects ProductCardDto then serializes to Redis with 10-minute TTL.
• All pods share same cache key space.

Practice next

1. Run Redis locally via Docker for dev.
2. Register IDistributedCache with StackExchangeRedis.
3. Cache DTO JSON not EF entities.
4. Publish cache bust message on ProductUpdated event.
5. Compare MessagePack serialization size vs JSON.

Remember

Redis shares cache across servers. EF runs only on cache miss. TTL and invalidation must match commerce rules.

ShopNest multi-pod catalog

Twelve API replicas share Redis product card cache populated by EF on miss.

Outcome: Database read QPS flat during viral product moment.