-
Notifications
You must be signed in to change notification settings - Fork 4
More About GraphSchema.io
Here's some more info about the GraphSchema.io project.
GraphSchema.io is built to support automation. So you can automate Dgraph and GraphSchema for your development, testing and production infrastructure.
The GraphSchema.io GraphQL API is mostly intended to be used from automated deployment pipelines and automated testing infrastructure, where you'd deploy, manage, backup, restore and destroy Dgraph or GraphSchema instances as part of an automated pipeline.
A Web-UI is also on the way with the beta release (expected in May), so with that you'll get a comfortable Web interface to manage your infrastructure and pet instances as well. But for now it's all about automation.
I'm the first customer of GraphSchema.io.
I've previously contributed a C# client library to the Dgraph community. Recently, I've been improving it. As part of it's automated end-to-end testing, it runs a suite of tests against a live Dgraph instance. In my cloud build pipelines (and my local development) that works by requesting GraphSchema.io to deploy a Dgraph instance, running the end-to-end testing against that cloud-hosted instance, and then tearing down the instance.
There's no prizes for guessing how I automate the testing of GraphSchema (which needs Dgraph instances), or how I plan to automate performances tests, etc for GraphSchema (which will need to deploy Dgraph+GraphSchema).
As well as automated testing, you could imagine doing a similar thing in a deployment pipeline to spin up your Dgraph or GraphSchema infrastructure.
Currently GraphSchema.io is in alpha testing with a small group accessing it to get free instances.
Because it's an alpha, don't expect these instances to be high spec. There's also no uptime or service level promises being made. The alpha isn't a performance test, it's about me getting to understand what features are needed, and about the alpha customers getting some free instances in return :-)
Once I feel that the alpha customers are happy with what's on offer, it'll move into a wider beta release with service tiers that'll offer higher spec machines, SSD, etc.
...no seriously, I said alpha. I can't guarantee that a development improvement won't cause me to blow away your instance - that doesn't really happen though, cause a redeployment of GraphSchema.io doesn't touch running instances, but still: it's not production, don't store important data here, etc., etc.
GraphSchema.io is a GraphQL API built using GraphSchema.
That means GraphSchema.io is a really thin GraphQL layer backed by a GraphSchema store (which is in turn backed by Dgraph). Because GraphSchema can do so much of the work of building a GraphQL API, GraphSchema.io turns out to be a GraphQL schema file defining the backing store and a really thin authentication and business logic layer. GraphSchema gives you loads of power for little work!
Watch out for rolling updates of GraphSchema with all the features that make this possible.
The Dgraph/GraphSchema instances themselves are hosted in a GCP kubernetes cluster
I'm using GraphSchema.io constantly.
I use it for development and testing of GraphSchema and the Dgraph C# client library.
As I need features to help me do that, I'm adding them to GraphSchema.io. I rely on it because I need it to run smoothly and work well with my deployment and testing pipelines. So I'm willing to support it and make GraphSchema.io and GraphSchema awesome!
I'd love other people to use it to!
I'm intending to run the free alpha with limited users for a month or so (with Dgraph instances starting now, and GraphSchema instances starting mid-late April). After that, if all looks good, I'll open up a beta phase to more users, but I'd have to charge a cost to cover the hosting costs. Beyond that, if enough people love it a much as I do, I'd be happy to open up more features, machine options, etc., as required by those using it.
There's other bit's in development, but not opened up in the alpha. For example:
- taking backups
- restoring from backups
- starting from uploaded data
I'll be working on any issues from the alpha and opening those features up to the alpha when they can be self-serviced easily.