{"id":17369,"date":"2025-07-25T10:09:07","date_gmt":"2025-07-25T17:09:07","guid":{"rendered":"https:\/\/www.couchbase.com\/blog\/?p=17369"},"modified":"2025-08-13T06:52:59","modified_gmt":"2025-08-13T13:52:59","slug":"choosing-the-right-query-language-sql-vs-mongo","status":"publish","type":"post","link":"https:\/\/www.couchbase.com\/blog\/choosing-the-right-query-language-sql-vs-mongo\/","title":{"rendered":"Choosing the Right Query Language: SQL++ vs. Mongo"},"content":{"rendered":"

When you\u2019re building apps, the query language you work with makes a big difference. It affects everything from performance to how quickly your team can ship features. In this post, I\u2019ve asked an LLM to create a list of 21 ranked criteria that are most valuable when deciding which query language to use. We\u2019ll walk through each one and show how SQL++, used by Couchbase, and Mongo\u2019s Query API (previously known as MQL) languages stack up.<\/p>\n

Whether you\u2019re evaluating databases for a new project, or just curious about the differences between these two approaches, this should give you a clear picture of where each one shines and where it might fall short.<\/p>\n

Expressiveness<\/h2>\n

SQL++ lets you write rich, complex queries\u2014joins, subqueries, aggregates, window functions\u2014while working with flexible document data. You can express nearly anything you\u2019d expect in SQL, plus native support for JSON.<\/p>\n

Mongo\u2019s query language is more limited. You can do a lot with the aggregation pipeline, but expressing multi-collection joins or deep subqueries can become verbose and may require workarounds or advanced pipeline constructs.<\/p>\n\n\n\n\n
SQL++<\/strong><\/th>\nMongo<\/strong><\/th>\n<\/tr>\n
\n
SELECT c.name, o.total\r\nFROM customers c\r\nJOIN orders o ON c.id = o.customer_id\r\nWHERE o.total > 100\r\nORDER BY o.total DESC;<\/pre>\n<\/td>\n
Mongo requires a more procedural pipeline:<\/p>\n
db.orders.aggregate([\r\n\u00a0 { $match: { total: { $gt: 100 } } },\r\n\u00a0 { $lookup: {\r\n\u00a0 \u00a0 \u00a0 from: \"customers\",\r\n\u00a0 \u00a0 \u00a0 localField: \"customer_id\",\r\n\u00a0 \u00a0 \u00a0 foreignField: \"id\",\r\n\u00a0 \u00a0 \u00a0 as: \"customer\"\r\n\u00a0 \u00a0 }\r\n\u00a0 },\r\n\u00a0 { $unwind: \"$customer\" },\r\n\u00a0 { $project: { name: \"$customer.name\", total: 1 } },\r\n\u00a0 { $sort: { total: -1 } }\r\n]);\r\n<\/pre>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n
Readability<\/h2>\n
SQL++ is declarative and familiar to anyone with SQL experience. The intent of the query is easy to see.<\/p>\n
SELECT name, age\r\nFROM users\r\nWHERE age >= 21\r\nORDER BY name;\r\n<\/pre>\n
Mongo\u2019s pipeline is more verbose and procedural\u2014you have to mentally step through each stage.<\/p>\n
db.users.find(\r\n\u00a0 { age: { $gte: 21 } },\r\n\u00a0 { name: 1, age: 1, _id: 0 }\r\n).sort({ name: 1 });\r\n<\/pre>\n
For simple queries this isn\u2019t terrible<\/b>. But as queries grow (joins, subqueries, aggregations), readability can drop quickly in Mongo\u2019s pipeline format.<\/p>\n
Consistency<\/h2>\n
SQL++ is highly consistent. The syntax and semantics are predictable, and it feels like “SQL for documents” across the board.<\/p>\n
Example with nested data:<\/p>\n
SELECT u.name, a.city\r\nFROM users u\r\nUNNEST u.addresses AS a\r\nWHERE a.country = \"USA\";\r\n<\/pre>\n
In Mongo, querying nested structures often requires switching to dot notation or array operators, and pipelines use a different style than find()<\/em>. You end up learning multiple syntaxes:<\/p>\n
Simple query with dot notation:<\/p>\n
db.users.find(\r\n\u00a0 { \"addresses.country\": \"USA\" },\r\n\u00a0 { name: 1, \"addresses.city\": 1, _id: 0 }\r\n);\r\n<\/pre>\n
More complex? You\u2019re back in aggregation-land:<\/p>\n
db.users.aggregate([\r\n\u00a0 { $unwind: \"$addresses\" },\r\n\u00a0 { $match: { \"addresses.country\": \"USA\" } },\r\n\u00a0 { $project: { name: 1, city: \"$addresses.city\", _id: 0 } }\r\n]);\r\n<\/pre>\n
Performance<\/h2>\n
When it comes to raw performance, Couchbase SQL++ consistently outpaces MongoDB\u2019s query language in benchmarks, especially under high-throughput, low-latency workloads.<\/p>\n
Couchbase has demonstrated superior scalability and efficiency in various benchmarks. For instance, in the Yahoo! Cloud Serving Benchmark (YCSB), Couchbase exhibited significantly better performance at scale\u2014up<\/a> compared to MongoDB.<\/p>\n
Beyond benchmarks, Couchbase customers have reported significant performance improvements:<\/p>\n