optimizing-ef-core-queries

Optimizing EF Core Queries

Diagnose and fix slow Entity Framework Core (EF Core) queries. Start from the generated SQL/logs, apply the smallest change that removes the bottleneck, and confirm the fix by re-reading the SQL and the query count. Prefer changes that reduce round-trips, duplicated rows, scans, or per-call translation cost over micro-optimizations. Apply one change at a time and re-measure.

When to Use

• EF Core queries are slow or emit far more SQL statements than expected
• The same query repeats once per row (N+1 / lazy loading)
• Multiple collection Includes blow up or duplicate rows
• Deep pages slow down as Skip grows, or bulk updates load rows just to modify them
• A filtered/sorted query scans even though the column is indexed, or a filtered/sorted column has no supporting index
• A hot, frequently-executed query pays EF Core's LINQ-translation cost on every call

When Not to Use

• The code uses Dapper or raw ADO.NET, not EF Core. Answer the SQL/indexing/query-plan question directly; do not introduce a DbContext or recommend AsNoTracking, Include, AsSplitQuery, or other EF Core APIs.

First: capture the generated SQL

You cannot optimize what you cannot see. Turn on command logging and read the SQL and query count before changing anything:

optionsBuilder.LogTo(Console.WriteLine, LogLevel.Information);
// or set "Microsoft.EntityFrameworkCore.Database.Command": "Information" in appsettings.json

Tag a query with .TagWith("...") to find it in the log. Count how many statements a slow operation runs, and how many rows each returns, before and after each change.

Fixes

Keep predicates sargable — never wrap an indexed column in a function

An index can only be used when the indexed column appears bare on one side of the comparison. Wrapping it in a function or arithmetic — CreatedAt.Year == y, CreatedAt.Date == d, ToLower(Name) == n, Price * 1.1 > x, or a leading-wildcard LIKE '%foo' — forces a per-row computation the index cannot satisfy, so the query scans the whole table even though the index exists. Adding another index changes nothing. Rewrite the predicate so the column stays bare, usually as a half-open range:

// Non-sargable: a function is computed for every row → full scan
db.Logs.Where(l

The same rule covers several common shapes:

• Case-insensitive text — compare a stored normalized column instead of ToLower(...)/ToUpper(...).
• Computed expressions — compare against the precomputed constant, not column * k > x.
• Converting the column to another type — a predicate over column.ToString() (for example matching the text form of a number or date, total.ToString().StartsWith(p)) applies a function to every row and often can't be translated to SQL at all, forcing a client-side evaluation that pulls the whole table into memory. Filter on the typed column with a real comparison or range instead.
• Substring search — name.Contains(term) becomes an unanchored LIKE '%term%' that can't seek an index and scans the table; a trailing-wildcard prefix (name.StartsWith(term) → 'term%') can seek. Anchor the search when a prefix match is acceptable — this changes which rows match, so confirm the behavior first — and put real substring or fuzzy search behind a full-text index on large tables.

Verify: the plan shows a seek/index instead of a scan and duration drops. If the column genuinely has no index, add one (see below) — but only after the predicate is sargable.

Compile hot, frequently-executed queries

On a very hot path that runs the same query shape thousands of times over a reused context, EF Core re-parses the LINQ expression tree and probes its query cache on every call. When the query is already minimal (an indexed lookup or a small projection) and read-only tweaks such as AsNoTracking buy nothing, that per-call translation is the remaining cost. Compile the query once with EF.CompileQuery / EF.CompileAsyncQuery and reuse the delegate:

The delegate is static (compiled once) and takes the DbContext plus each parameter as arguments. Use it for endpoints or loops that execute one query shape at very high frequency; it does nothing for one-off queries.

Verify: the hot loop's mean time drops with identical results.

Remove N+1 and avoid lazy loading

The same SELECT repeated once per row (a navigation accessed inside a loop) is an N+1. Load the related data in one round-trip — project the aggregates with Select, or eager-load with Include:

Prefer projection or Include over lazy loading: lazy loading is a leading cause of N+1 and forces synchronous I/O. In server apps, don't enable Microsoft.EntityFrameworkCore.Proxies or mark navigations virtual for lazy loading.

Verify: a fixed, small query count regardless of row count.

Split multiple collection Includes

Includeing two or more collection navigations in one query multiplies rows (a Cartesian explosion) and duplicates parent data. Use AsSplitQuery() so each collection loads in its own statement; add OrderBy on a unique key so rows stitch together:

Verify: rows per statement drop sharply and total duration improves.

Filter and paginate; prefer keyset over offset

Constrain large result sets with Where, and page with keyset (seek) pagination rather than Skip/Take, which still scans and discards the skipped rows on deep pages:

Order by a unique, stable, indexed key (add tie-breakers if the sort column isn't unique). Keyset pages by the last key seen rather than a page number, so it changes the method's inputs; when a fixed signature rules out an in-place switch, still flag the deep-offset scan and recommend keyset.

Verify: page latency stays roughly constant from early to deep pages.

Add missing indexes

Moving a filter into SQL or making a predicate sargable stops the client-side waste, but a WHERE or ORDER BY on a column with no index still scans the whole table inside the database — and a frequently-run query then re-scans it on every call. So audit index coverage separately from the query rewrite: for each query, check whether its filter and sort columns are backed by an index. Entity keys and foreign keys are indexed by convention, but other columns — status flags, state/enum fields, timestamps, names — usually are not unless the model configures it. When a hot query filters or sorts on such an unindexed column, recommend adding an index and say so explicitly, even when the rewritten query already returns the right rows: the index is a separate fix the code change alone doesn't deliver. (If the predicate isn't sargable, fix that first — a new index can't help a scan caused by a function on the column.)

If EF Core owns the schema, add the index in the model and migrate:

Then create the migration with dotnet ef migrations add .... Do not apply it with dotnet ef database update (or any equivalent that writes to the database) without explicit user approval — applying a migration mutates the database, so add the migration, show it to the user, and let them run the update once they've reviewed it. If EF Core does not own the schema, recommend the same index to whoever manages the database. Don't over-index — every index slows writes.

Verify: the plan uses a seek/index instead of a scan.

Set-based bulk updates and deletes

Replace a load-mutate-SaveChanges loop with ExecuteUpdateAsync/ExecuteDeleteAsync (EF Core 7+) — one statement, no entities materialized:

These bypass the change tracker and EF-side cascade behavior — apply related changes explicitly.

Verify: a single UPDATE/DELETE with a WHERE and no preceding SELECT.

Common Pitfalls

References

• Efficient querying — EF Core
• Compiled queries — EF Core
• Efficient updating (ExecuteUpdate/ExecuteDelete) — EF Core
• Single vs. split queries
• Pagination
• Indexes