此页面由 Cloud Translation API 翻译。

优化对多个属性使用范围和不等式过滤条件的查询

本页面提供了一些索引编制策略的示例，可在对多个字段使用范围和不等式过滤条件的查询中使用这些策略来打造高效的查询体验。

使用查询解释优化查询

如需确定所使用的查询和索引是否最优，您可以使用查询解释创建查询，然后查看执行摘要。

Java

...
// Build the query
Query<Entity> query =
    Query.newEntityQueryBuilder()
        .setKind("employees")
        .setFilter(
            CompositeFilter.and(
                PropertyFilter.gt("salary", 100000), PropertyFilter.gt("experience", 0)))
        .setOrderBy(OrderBy("experience"), OrderBy("salary"))
        .build();

// Set the explain options to get back *only* the plan summary
ExplainResults<Entity> explainResults = datastore.run(query, ExplainOptions.newBuilder().build());

// Get the explain metrics
Optional<ExplainMetrics> explainMetrics = results.getExplainMetrics();
if (!explainMetrics.isPresent()) {
  throw new Exception("No explain metrics returned");
}

// Get the plan summary
PlanSummary planSummary = explainMetrics.get().getPlanSummary();
List<Map<String, Object>> indexesUsed = planSummary.getIndexesUsed();
System.out.println("----- Indexes Used -----");
indexesUsed.forEach(map -> map.forEach((s, o) -> System.out.println(s + ": " + o)));

// Get the execution stats
if (!explainMetrics.getExecutionStats().isPresent()) {
  throw new Exception("No execution stats returned");
}

ExecutionStats queryStats = explainMetrics.getExecutionStats().get();
Map<String, Object> debugStats = queryStats.getDebugStats();
System.out.println("----- Debug Stats -----");
debugStats.forEach((s, o) -> System.out.println(s + ": " + o));

以下示例展示了如何使用正确的索引顺序来减少 Datastore 模式 Firestore 扫描的实体数量。

简单查询

在前面的雇员集合示例中，使用 (salary, experience) 索引运行的简单查询如下所示：

GQL

SELECT *
FROM /employees
WHERE salary > 100000 AND experience > 0
ORDER BY experience, salary;

Java

Query<Entity> query =
   Query.newEntityQueryBuilder()
    .setKind("employees")
    .setFilter(
        CompositeFilter.and(
            PropertyFilter.gt("salary", 100000), PropertyFilter.gt("experience", 0)))
    .setOrderBy(OrderBy("experience"), OrderBy("salary"))
    .build();

该查询扫描了 95,000 个索引条目，仅返回了 5 个实体。虽然系统读取了大量索引条目，但由于它们不满足查询谓词，因此被滤除。

// Output query planning info
{
        "indexesUsed": [
            {
                "query_scope": "Collection Group",
                "properties": "(experience ASC, salary ASC, __name__ ASC)"
            }
        ]
    },
    // Output Query Execution Stats
    {
        "resultsReturned": "5",
        "executionDuration": "2.5s",
        "readOperations": "100",
        "debugStats": {
            "index_entries_scanned": "95000",
            "documents_scanned": "5",
            "billing_details": {
                "documents_billable": "5",
                "index_entries_billable": "95000",
                "small_ops": "0",
                "min_query_cost": "0"
            }
        }
    }

根据前面的示例，我们可以推断出 salary 约束条件比 experience 约束条件更为严苛。

GQL

SELECT *
FROM /employees
WHERE salary > 100000 AND experience > 0
ORDER BY salary, experience;

Java

Query<Entity> query =
   Query.newEntityQueryBuilder()
    .setKind("employees")
    .setFilter(
        CompositeFilter.and(
            PropertyFilter.gt("salary", 100000), PropertyFilter.gt("experience", 0)))
    .setOrderBy(OrderBy("salary"), OrderBy("experience"))
    .build();

如果您明确使用 orderBy() 子句按上述顺序添加谓词，Datastore 模式 Firestore 会使用 (salary, experience) 索引来运行查询。由于第一个范围过滤条件的选择比之前的查询更好，因此该查询运行速度更快、更经济高效。

    // Output query planning info
{
    "indexesUsed": [
        {
            "query_scope": "Collection Group",
            "properties": "(salary ASC, experience ASC, __name__ ASC)"
        }
    ],
    // Output Query Execution Stats
        "resultsReturned": "5",
        "executionDuration": "0.2s",
        "readOperations": "6",
        "debugStats": {
            "index_entries_scanned": "1000",
            "documents_scanned": "5",
            "billing_details": {
                "documents_billable": "5",
                "index_entries_billable": "1000",
                "small_ops": "0",
                "min_query_cost": "0"
            }
        }
    }

后续步骤

了解查询解释。