Here we test what an impact can Grapher have on the performance and cost of your infrastructure, provided you use MongoDB and you make use of a lot of relations.
Quickly create your own cluster to test this out: http://mbsy.co/lHRhH it's free
Summary:
| Query | Grapher | Standard | Improvement |
|---|---|---|---|
| 1 | 256ms | 1040ms | 4x |
| 2 | 302ms | 10650ms | 35x |
| 3 | 559ms | >120s | 200x |
Verdict: INCREDIBLY EXPONENTIAL.
Look in TEST RESULTS to see how we did the tests, so you can do them too!
We use:
- cultofcoders:apollo
- cultofcoders:grapher
meteor npm i
meteor runTo do justice to Grapher:
- Use a remote database
- We should have had many other fields in db, not only the minimal ones, that's how it would have shined even more!
- server/load.js - Here you can enable/disable Grapher
- server/entities/* - Here are the GraphqlTypes
- imports/fixtures - Here is where the demo data is loaded
- private/visualise.html - Here you can visualise the data schema
- queries/QueryGrapher.resolver.js - This is how Grapher resolvers look like
- queries/QueryNormal.resolver.js + queries/resolvers/* - The resolvers needed for non Grapher linking
Fixtures will start running by default look in server/imports/fixtures.js
Open private/visualise.html in your browser, while Meteor is running
db.users.remove({});
db.comments.remove({});
db.tags.remove({});
db.groups.remove({});
db.posts.remove({});
db.postCategories.remove({});