5 ms·

One of the most important things I learned with databases was to run each of my queries using EXPLAIN (EXPLAIN QUERY PLAN in sqlite) and seeing which indexes ar

One of the most important things I learned with databases was to run each of my queries using EXPLAIN (EXPLAIN QUERY PLAN in sqlite) and seeing which indexes are being used, if any. One of the reasons I don't like ORMs is that I'm not able to see the underlining query and truly optimize a service. That may be fine for a new service where performance isn't crucial, but once it needs to scale, you need to put on your engineering hat, get your hands dirty, and optimize queries. You'll find you need to re-write queries so that there isn't complex nesting in the WHERE statement and flatten your logic so that the SQL optimizer can use your indexes. You may need to put SELECT statements within SELECT statements, where the innermost SELECT uses indexes and the outer queries are using the result of the inner query, which is smaller than the whole table.

obi1kenobi 6y ago
I wrote a thread on this on Twitter: https://twitter.com/PredragGruevski/status/1263916599062540288?s=20 https://twitter.com/PredragGruevski/status/12639165990625402... I feel that SQL aimed to be Python and became x86 assembly instead. It's no longer a simple "just works" query language the moment you have to worry about predicate flattening, join decomposition, CTEs that introduce optimization barriers, and "IN()" being faster than equivalent "JOINs". As a result, I started a project that allows you to write read-only database-agnostic queries called GraphQL compiler: https://graphql-compiler.readthedocs.io/ https://graphql-compiler.readthedocs.io/ https://github.com/kensho-technologies/graphql-compiler https://github.com/kensho-technologies/graphql-compiler The core idea of the project is to get us the convenience of specifying the "what question I want answered," but without the inconvenience of "how is the answer computed / with which specific set of queries / where did the data come from?" -- unless you want to peek under the hood, of course. All the visibility into the nitty-gritty details available on demand, but without the tedium of having to hand-optimize queries and know all the "magic" ways in which queries get faster or slower for each individual kind of database.
- petergeoghegan 6y ago
  > The core idea of the project is to get us the convenience of specifying the "what question I want answered," but without the inconvenience of "how is the answer computed / with which specific set of queries / where did the data come from?" So...exactly like SQL, then?
  - ghusbands 6y ago
    The post you're replying to directly addresses that. When you write SQL: > It's no longer a simple "just works" query language the moment you have to worry about predicate flattening, join decomposition, CTEs that introduce optimization barriers, and "IN()" being faster than equivalent "JOINs". Though it doesn't seem to address how to optimize things using the GraphQL compiler, when there's a need, without massaging the queries, as with SQL.
    - random_savv 6y ago
      I think the parent's point might have been: SQL tried with mixed success. What makes the GraphQL compiler different?
    - obi1kenobi 6y ago
      Just like how GCC and Clang/LLVM know all the quirks of various CPUs and can optimize accordingly, GraphQL compiler aims to know the quirks of various databases (down to individual database versions: e.g., in Postgres 12 certain kinds of CTEs are no longer an optimization barrier) and optimize accordingly. This is clearly a massive challenge, but one made easier by the fact that GraphQL compiler queries (unlike SQL queries) operate at a much higher level of abstraction. In SQL, you write "here's a CTE, now recursively JOIN X.foo to Y.bar" where X and Y could be just about anything, even something where a recursive JOIN is nonsensical. If you want to put a WHERE clause, you have to decide whether it goes in the recursive CTE itself, in a separate CTE that is ordered before the recursive CTE, or if you want to wrap the recursive CTE into another SELECT and put the WHERE there. The correct answer varies from database to database, and as a function of the size, layout, and index coverage of your data. In GraphQL compiler, your queries are much more declarative in comparison: your query would say "find all subgraphs where vertex A's field 'foo' has value 123, and where A has a recursively-expanded edge (i.e. 0+ hops along that edge) to a vertex with field 'bar' with value 456". It's then the compiler's job to figure out which of the many equivalent SQL statements (or other db language queries, if you aren't using SQL) is going to be the best way to compute the result you asked for. Here's an example from our test suite: input query: https://github.com/kensho-technologies/graphql-compiler/blob/main/graphql_compiler/tests/test_input_data.py#L941 https://github.com/kensho-technologies/graphql-compiler/blob... Microsoft SQL Server-flavored compiled SQL output: https://github.com/kensho-technologies/graphql-compiler/blob/main/graphql_compiler/tests/test_compiler.py#L2449 https://github.com/kensho-technologies/graphql-compiler/blob... I'm writing a blog post about this with more detail, follow me on Twitter if you'd like to see it when it comes out.
- Someone 6y ago
  It’s a leaky abstraction, yes. In the ideal world, SQL query planners always would find the optimal plan in zero seconds and DDL would allow you to express non-functional requirements for various kinds of queries (“retrieving this record by ID should take at most 1ms”, “looking up related records in that table by order number should…”), but the world isn’t ideal. We have to manually create indices, split tables, move different tables to different disks, etc. for performance. “The GraphQL compiler turns read-only queries written in GraphQL syntax to different query languages” is useful for GraphQL ‘fans’, but I don’t see how that is going to solve that. Designing a user friendly query language for relational data isn’t the hard part. Executing such queries efficiently is. For SQL, there’s half a century of research on that. This paper is part of it, and indicates that, at this moment in time, effort is better spent on methods for keeping statistics on data up to date than on making cost models more fine grained.
  - obi1kenobi 6y ago
    > Designing a user friendly query language for relational data isn’t the hard part. Executing such queries efficiently is. I couldn't agree more with this part :) To be clear, GraphQL compiler isn't trying to "solve" SQL itself, just merely the fact that by the time all the table splitting, sharding across disks and machines, and similar required maintenance operations are done, your SQL query has grown impractically complex and entirely unreasonable to write. At sufficient scale, and with sufficient additional non-SQL databases in play, writing adequate queries becomes wildly impractical. In the GraphQL compiler world, all your databases (SQL and non-SQL) are represented in one unified schema against which you write database-agnostic queries, and GraphQL compiler handles the nitty-gritty details of "which query runs where." GraphQL compiler is not a toy project I build for fun — it's a core piece of data infrastructure that Kensho (company where I work) has been happily using in production for over 3 years now. I'm writing a blog post about exactly this, and I hope to publish it very soon! Follow me on Twitter if you'd like to see it when it comes out.
vsareto 6y ago
>One of the reasons I don't like ORMs is that I'm not able to see the underlining query and truly optimize a service. I thought most of them had some feature where you could dump the query before it gets sent to the DB. Stuff like this: https://stackoverflow.com/questions/1412863/how-do-i-view-the-sql-generated-by-the-entity-framework https://stackoverflow.com/questions/1412863/how-do-i-view-th...
- devmunchies 6y ago
  yeah but if the generated SQL doesn't look how you want it, now you gotta optimize it using the ORM's language and not SQL. Tweak the ORM, see the generated SQL, tweak again... etc. If you're looking at the generated SQL I would rather just use the SQL directly in my code. There's probably features in ORMs where you can write raw SQL and tell it how to map the result to an object but I haven't used an ORM in a while.
  - jacobsenscott 6y ago
    Generally it is the table design, and not the ORM that is the constraint on how you can write your queries. A good ORM lets you customize any part of the query, or even just write plain sql. Most performance problems that pop up as a product matures aren't because the ORM generates "slow" queries, it is because the table design didn't scale. That can't be fixed by writing plain sql.
    - devmunchies 6y ago
      This doesn’t really make sense to me. You don’t just design a good table schema independently and start querying it. The design isn’t a step that comes before the queries. The queries and the design are created in tandem. You design the schemas with the indices and queries in mind, write queries that use your indices. A good design doesn’t magically scale. It’s based on how you set up your indices and write SQL. A good table design is only good BECAUSE it enables efficient predicate use on the SQL queries. You can’t just query any column willy nilly, you have to plan it. That’s why I like thinking in SQL with the table definition on-hand. Example, if I write “SELECT * WHERE x OR y” and “y” isn’t indexed, then this will do a full table scan. Not ok. I need to plan my queries so it does something like “WHERE x OR (y AND z)” where “z” is indexed so it filters by “z” and then “y”. I don’t want to have to try and figure out how to get the ORM to produce that.
ris 6y ago
> One of the reasons I don't like ORMs is that I'm not able to see the underlining query and truly optimize a service. The vast majority of your queries will not fall into the category of "bottlenecks that need to be optimized" though, and you (and your probably more inexperienced team) will benefit massively from the less error-prone & more extensible nature of ORMs (I never again want to have to deal with an attempt at SQL code reuse that has grown into a string-formatted, quadruple-manifestation, triple-escaped nightmare) A good ORM will also ease the transition into more manual SQL too, so that you can still retain the benefits of e.g. uniform abstract objects app-side.
lmm 6y ago
This kind of thing is the reasons why I do like ORMs. Even if you can see the SQL, that doesn't mean you can tell what it's doing: you don't know what indices will be used or not used, you don't know why the query planner will do things one way or another, you're not actually "close to the metal" at all. So might as well have the convenience of an ORM; sure, you'll have to do some fiddly profiling as and when you have performance problems, but even if you were using hand-written SQL you'd still have to do that.
- devmunchies 6y ago
  > Even if you can see the SQL, that doesn't mean you can tell what it's doing You can. I mentioned EXPLAIN in my comment. And the query planner isn’t a black box. Once you read the documentation on how the order of operations is determined by the engine, you can start to be thinking on the same plane as the query engine. You can infer how a query will use indices and the way the WHERE clause will be used. Admittedly it’s not as easy as using an ORM, but if you’re a SQL expert then you can make queries much more optimized. You’ll never internalize how a SQL interpreter reads your queries unless you do it. I’m talking about huge tables that are hit many times a second, where you need to start thinking like a Formula One team, being creative with queries to shave off hundreds of milliseconds.
  - lmm 6y ago
    > And the query planner isn’t a black box. Once you read the documentation on how the order of operations is determined by the engine, you can start to be thinking on the same plane as the query engine. You can infer how a query will use indices and the way the WHERE clause will be used. If you're willing to put that kind of time and effort in, you should have no trouble understanding how your ORM generates queries - IME they tend to be far clearer, better documented, and more introspectable than database query planners.
    - devmunchies 6y ago
      It’s not that I’m willing to put in the effort, it’s that the scale I operate at requires it. I would rather become highly proficient at SQL and use it universally with any server language than become an expert at a specific ORM. But I don’t see how one could become an expert at writing ORM queries without knowing the underlining SQL, which means you’re putting in the time to master two languages. I may be just be an outlier, and that’s fine. I like geeking out over SQL optimization.
rst 6y ago
That depends on the ORM -- ActiveRecord has a ".explain" method on the proxy objects for its query builder which displays the SQL, and the query plan (displayed in db-specific format).
- why-el 6y ago
  Note that AR's `explain` actually runs the query (effectively an explain analyze). This might or might not matter depends on what you are trying to achieve.

obi1kenobi 6y ago

I wrote a thread on this on Twitter: https://twitter.com/PredragGruevski/status/1263916599062540288?s=20 https://twitter.com/PredragGruevski/status/12639165990625402... I feel that SQL aimed to be Python and became x86 assembly instead. It's no longer a simple "just works" query language the moment you have to worry about predicate flattening, join decomposition, CTEs that introduce optimization barriers, and "IN()" being faster than equivalent "JOINs". As a result, I started a project that allows you to write read-only database-agnostic queries called GraphQL compiler: https://graphql-compiler.readthedocs.io/ https://graphql-compiler.readthedocs.io/ https://github.com/kensho-technologies/graphql-compiler https://github.com/kensho-technologies/graphql-compiler The core idea of the project is to get us the convenience of specifying the "what question I want answered," but without the inconvenience of "how is the answer computed / with which specific set of queries / where did the data come from?" -- unless you want to peek under the hood, of course. All the visibility into the nitty-gritty details available on demand, but without the tedium of having to hand-optimize queries and know all the "magic" ways in which queries get faster or slower for each individual kind of database.

petergeoghegan 6y ago
> The core idea of the project is to get us the convenience of specifying the "what question I want answered," but without the inconvenience of "how is the answer computed / with which specific set of queries / where did the data come from?" So...exactly like SQL, then?
- ghusbands 6y ago
  The post you're replying to directly addresses that. When you write SQL: > It's no longer a simple "just works" query language the moment you have to worry about predicate flattening, join decomposition, CTEs that introduce optimization barriers, and "IN()" being faster than equivalent "JOINs". Though it doesn't seem to address how to optimize things using the GraphQL compiler, when there's a need, without massaging the queries, as with SQL.
  - random_savv 6y ago
    I think the parent's point might have been: SQL tried with mixed success. What makes the GraphQL compiler different?
  - obi1kenobi 6y ago
    Just like how GCC and Clang/LLVM know all the quirks of various CPUs and can optimize accordingly, GraphQL compiler aims to know the quirks of various databases (down to individual database versions: e.g., in Postgres 12 certain kinds of CTEs are no longer an optimization barrier) and optimize accordingly. This is clearly a massive challenge, but one made easier by the fact that GraphQL compiler queries (unlike SQL queries) operate at a much higher level of abstraction. In SQL, you write "here's a CTE, now recursively JOIN X.foo to Y.bar" where X and Y could be just about anything, even something where a recursive JOIN is nonsensical. If you want to put a WHERE clause, you have to decide whether it goes in the recursive CTE itself, in a separate CTE that is ordered before the recursive CTE, or if you want to wrap the recursive CTE into another SELECT and put the WHERE there. The correct answer varies from database to database, and as a function of the size, layout, and index coverage of your data. In GraphQL compiler, your queries are much more declarative in comparison: your query would say "find all subgraphs where vertex A's field 'foo' has value 123, and where A has a recursively-expanded edge (i.e. 0+ hops along that edge) to a vertex with field 'bar' with value 456". It's then the compiler's job to figure out which of the many equivalent SQL statements (or other db language queries, if you aren't using SQL) is going to be the best way to compute the result you asked for. Here's an example from our test suite: input query: https://github.com/kensho-technologies/graphql-compiler/blob/main/graphql_compiler/tests/test_input_data.py#L941 https://github.com/kensho-technologies/graphql-compiler/blob... Microsoft SQL Server-flavored compiled SQL output: https://github.com/kensho-technologies/graphql-compiler/blob/main/graphql_compiler/tests/test_compiler.py#L2449 https://github.com/kensho-technologies/graphql-compiler/blob... I'm writing a blog post about this with more detail, follow me on Twitter if you'd like to see it when it comes out.
Someone 6y ago
It’s a leaky abstraction, yes. In the ideal world, SQL query planners always would find the optimal plan in zero seconds and DDL would allow you to express non-functional requirements for various kinds of queries (“retrieving this record by ID should take at most 1ms”, “looking up related records in that table by order number should…”), but the world isn’t ideal. We have to manually create indices, split tables, move different tables to different disks, etc. for performance. “The GraphQL compiler turns read-only queries written in GraphQL syntax to different query languages” is useful for GraphQL ‘fans’, but I don’t see how that is going to solve that. Designing a user friendly query language for relational data isn’t the hard part. Executing such queries efficiently is. For SQL, there’s half a century of research on that. This paper is part of it, and indicates that, at this moment in time, effort is better spent on methods for keeping statistics on data up to date than on making cost models more fine grained.
- obi1kenobi 6y ago
  > Designing a user friendly query language for relational data isn’t the hard part. Executing such queries efficiently is. I couldn't agree more with this part :) To be clear, GraphQL compiler isn't trying to "solve" SQL itself, just merely the fact that by the time all the table splitting, sharding across disks and machines, and similar required maintenance operations are done, your SQL query has grown impractically complex and entirely unreasonable to write. At sufficient scale, and with sufficient additional non-SQL databases in play, writing adequate queries becomes wildly impractical. In the GraphQL compiler world, all your databases (SQL and non-SQL) are represented in one unified schema against which you write database-agnostic queries, and GraphQL compiler handles the nitty-gritty details of "which query runs where." GraphQL compiler is not a toy project I build for fun — it's a core piece of data infrastructure that Kensho (company where I work) has been happily using in production for over 3 years now. I'm writing a blog post about exactly this, and I hope to publish it very soon! Follow me on Twitter if you'd like to see it when it comes out.

vsareto 6y ago

>One of the reasons I don't like ORMs is that I'm not able to see the underlining query and truly optimize a service. I thought most of them had some feature where you could dump the query before it gets sent to the DB. Stuff like this: https://stackoverflow.com/questions/1412863/how-do-i-view-the-sql-generated-by-the-entity-framework https://stackoverflow.com/questions/1412863/how-do-i-view-th...

devmunchies 6y ago
yeah but if the generated SQL doesn't look how you want it, now you gotta optimize it using the ORM's language and not SQL. Tweak the ORM, see the generated SQL, tweak again... etc. If you're looking at the generated SQL I would rather just use the SQL directly in my code. There's probably features in ORMs where you can write raw SQL and tell it how to map the result to an object but I haven't used an ORM in a while.
- jacobsenscott 6y ago
  Generally it is the table design, and not the ORM that is the constraint on how you can write your queries. A good ORM lets you customize any part of the query, or even just write plain sql. Most performance problems that pop up as a product matures aren't because the ORM generates "slow" queries, it is because the table design didn't scale. That can't be fixed by writing plain sql.
  - devmunchies 6y ago
    This doesn’t really make sense to me. You don’t just design a good table schema independently and start querying it. The design isn’t a step that comes before the queries. The queries and the design are created in tandem. You design the schemas with the indices and queries in mind, write queries that use your indices. A good design doesn’t magically scale. It’s based on how you set up your indices and write SQL. A good table design is only good BECAUSE it enables efficient predicate use on the SQL queries. You can’t just query any column willy nilly, you have to plan it. That’s why I like thinking in SQL with the table definition on-hand. Example, if I write “SELECT * WHERE x OR y” and “y” isn’t indexed, then this will do a full table scan. Not ok. I need to plan my queries so it does something like “WHERE x OR (y AND z)” where “z” is indexed so it filters by “z” and then “y”. I don’t want to have to try and figure out how to get the ORM to produce that.

ris 6y ago

> One of the reasons I don't like ORMs is that I'm not able to see the underlining query and truly optimize a service. The vast majority of your queries will not fall into the category of "bottlenecks that need to be optimized" though, and you (and your probably more inexperienced team) will benefit massively from the less error-prone & more extensible nature of ORMs (I never again want to have to deal with an attempt at SQL code reuse that has grown into a string-formatted, quadruple-manifestation, triple-escaped nightmare) A good ORM will also ease the transition into more manual SQL too, so that you can still retain the benefits of e.g. uniform abstract objects app-side.

lmm 6y ago

This kind of thing is the reasons why I do like ORMs. Even if you can see the SQL, that doesn't mean you can tell what it's doing: you don't know what indices will be used or not used, you don't know why the query planner will do things one way or another, you're not actually "close to the metal" at all. So might as well have the convenience of an ORM; sure, you'll have to do some fiddly profiling as and when you have performance problems, but even if you were using hand-written SQL you'd still have to do that.

devmunchies 6y ago
> Even if you can see the SQL, that doesn't mean you can tell what it's doing You can. I mentioned EXPLAIN in my comment. And the query planner isn’t a black box. Once you read the documentation on how the order of operations is determined by the engine, you can start to be thinking on the same plane as the query engine. You can infer how a query will use indices and the way the WHERE clause will be used. Admittedly it’s not as easy as using an ORM, but if you’re a SQL expert then you can make queries much more optimized. You’ll never internalize how a SQL interpreter reads your queries unless you do it. I’m talking about huge tables that are hit many times a second, where you need to start thinking like a Formula One team, being creative with queries to shave off hundreds of milliseconds.
- lmm 6y ago
  > And the query planner isn’t a black box. Once you read the documentation on how the order of operations is determined by the engine, you can start to be thinking on the same plane as the query engine. You can infer how a query will use indices and the way the WHERE clause will be used. If you're willing to put that kind of time and effort in, you should have no trouble understanding how your ORM generates queries - IME they tend to be far clearer, better documented, and more introspectable than database query planners.
  - devmunchies 6y ago
    It’s not that I’m willing to put in the effort, it’s that the scale I operate at requires it. I would rather become highly proficient at SQL and use it universally with any server language than become an expert at a specific ORM. But I don’t see how one could become an expert at writing ORM queries without knowing the underlining SQL, which means you’re putting in the time to master two languages. I may be just be an outlier, and that’s fine. I like geeking out over SQL optimization.

rst 6y ago

That depends on the ORM -- ActiveRecord has a ".explain" method on the proxy objects for its query builder which displays the SQL, and the query plan (displayed in db-specific format).

why-el 6y ago
Note that AR's `explain` actually runs the query (effectively an explain analyze). This might or might not matter depends on what you are trying to achieve.