I'm getting the impression that if nothing else though, I'm probably gonna need to wait a few years for this tech to stabilize more. Which I find very unfortunate, because the standard way of doings things these days feels very unpleasant to me (I hate writing boilerplate more than anything: have an overuse injury, so typing is the worst part of coding)βand the GraphQL queries do seem to make a lot of sense (although forcing clients to have explicit knowledge of schema structure seems a little dangerous...).
If I have a schema like this:
users
projects (many per user)
blog_posts (many per user)
What SQL do you generate to handle a request for everything in a single query?simplified example query https://pastebin.com/7J2ZswRC
If you have a 3 level query (3 tables), the best you can hope for is to get 3 sequential queries, like get first level, collect the ids, request the second level, collect the ids, get 3rd level. It gets even more complicated teh more levels you and the bigger the dataset returned.
all of this can be done using a single join which is one roundtrip and it's faster.
In the 'join in SQL' case you could identify particular cases which are doing sequential queries and implement a different loader which just does one. It's not automatic but perhaps in most cases it's not necessary to do this step anyway. In the worst case you're back to doing as much work as you would for a bespoke API endpoint, but that's not the typical case. How much of a problem this ends up being in practice very much depends on the type of app you're building and how you intend to scale it.
In practice this probably means adding some logging of how long requests take and graphing it (say, 95th percentile request time) from time to time to spot pathological queries. Even better if you can automate it. I think this is stuff that everyone should be doing (after a certain stage), regardless of whether you are using GraphQL or a bespoke JSON API.
It's perfectly valid to identify subtrees of your query that would benefit from being executed as a database query, but to do that to your entire query just sounds like you're asking for trouble, I'd even go so far as to call it a premature optimization.
I am talking about queries/joins between tables that have foreign keys between them, like client/project/task/comment. I bet 90% of graphql schemas expose those kinds of relations between types.
For those type of relations (with FK) i can generate a single query that is as fast as it can be (certainly faster then dataloader) and as far as i've tested (a few millions of rows in tables, 3-7 levels in a query) i didn't leave the fast territory :) Of course there might be edge cases ...
About premature optimisations. Everyone likes to quote that, but never the full one which is "We should forget about small efficiencies, say about 97% of the time: premature optimization is the root of all evil." https://shreevatsa.wordpress.com/2008/05/16/premature-optimi...
The paper where it was published was about using goto to optimize loops and such small things, it was never "targeted" at algorithms and architecture.
As i think i mentioned here above, i was getting 10X throughput with joins compared to dataloader and i would not call that premature.
I get my data for a full UI within the budget i've allowed for uncached scenarios (100ms, but I want to go to 50ms). The approach you're suggesting will (in some circumstances) give me some short-terms win in terms of throughput and response times, but you've not said anything to suggest that I won't lose these benefits as I gradually transition into a domain-driven or micro-services (yuck) architecture.
I like to build my GraphQL servers under the assumption of a domain-driven architecture (because that's where all the projects I've worked on seem to end up, your mileage may vary), and then shoe-horn in some short-term performance tricks when I can.
I'm possibly a special snowflake here, but it's been a long time since i've had the opportunity to work on a project where I can go straight to the DB. Be it Elastic Search, a 3rd party, ill-advised micro-services, or complex logic in-between storage and presentation; nothing has quite been a pure DB project in the last 6-7 years.
Of course, you could argue this is premature architecture ;) but many of these complexities are from day one, or at least pretty early in a project's life.
I've spent a significant amount of time over the last year or so optimizing GraphQL servers built on domain-driven services (i.e. joins aren't an option) and managed to get to equal (or very marginally worse) performance to existing handcrafted endpoints that returned equivalent data (it was possible to build the same UI, even though the payloads weren't identical).
There are areas where GraphQL is inherently inefficient (trying to work on ways to mitigate these issues), but the reality is that deeply-nested UI appears to be less of a problem than I originally thought it would be.
You need to be very careful. To be optimal, you'll basically need to write very complex resolvers that inspect the AST themselves and fetch the data optimally which at that point is almost as writing your custom execution module.
That is basically what i did, custom execution module to translate a graphql request to a single sql query (https://subzero.cloud/)