runtime

README

runtime

The runtime a data infrastructure proxy and orchestrator – our data plane. It connects to data infrastructure and is or will be responsible for transpiling queries, parsing code artifacts, reconciling infra state, implementing connectors, enforcing (row-based) access policies, scheduling tasks, triggering alerts, and much more.

It's designed as a modular component that can be embedded in local applications (as it is into Rill Developer) or deployed stand-alone in a cloud environment.

Code structure

• cmd contains a main.go file that starts the runtime as a standalone server.
• connectors contains connector implementations.
• drivers contains interfaces and drivers for external data infrastructure that the runtime interfaces with (like DuckDB and Druid).
• pkg contains utility libraries.
• queries contains pre-defined analytical queries that the runtime can serve (used for profiling and dashboards).
• server contains a server that implements the runtime's APIs.
• sql contains bindings for the SQL native library (see the sql folder at the repo root for details).
• testruntime contains helper functions for initializing a test runtime with test data.

How to test and run

You can run and test the runtime as any other Go application. Start the server using:

go run ./runtime/cmd

Or run all tests using:

go test ./runtime/...

Configuration

The runtime server is configured using environment variables parsed in runtime/cmd/main.go. All environment variables have reasonable defaults suitable for local development. The current defaults are:

RILL_RUNTIME_ENV="development"
RILL_RUNTIME_HTTP_PORT="8080"
RILL_RUNTIME_GRPC_PORT="9090"
RILL_RUNTIME_LOG_LEVEL="info"
RILL_RUNTIME_DATABASE_DRIVER="sqlite"
RILL_RUNTIME_DATABASE_URL=":memory:"

Adding a new endpoint

We define our APIs using gRPC and use gRPC-Gateway to map the RPCs to a RESTful API. See proto/README.md for details.

To add a new endpoint:

1. Describe the endpoint in proto/rill/runtime/v1/api.proto
2. Re-generate gRPC and OpenAPI interfaces by running make proto.generate
3. Copy the new handler signature from the RuntimeServiceServer interface in proto/gen/rill/runtime/v1/api_grpc_pb.go
4. Paste the handler signature and implement it in a relevant file in runtime/server/

Adding a new analytical query endpoint

1. Add a new endpoint for the query by following the steps in the section above ("Adding a new endpoint")
2. Implement the query in runtime/queries by following the instructions in runtime/queries/README.md

Example: Creating an instance and rehydrating from code artifacts

# Start runtime
go run ./runtime/cmd/main.go

# Create instance
curl --request POST --url http://localhost:8080/v1/instances --header 'Content-Type: application/json' \
  --data '{
    "instance_id": "default",
    "olap_driver": "duckdb",
    "olap_dsn": "test.db",
    "repo_driver": "file",
    "repo_dsn": "./examples/ad_bids",
    "embed_catalog": true
}'

# Apply code artifacts
curl --request POST --url http://localhost:8080/v1/instances/default/reconcile --header 'Content-Type: application/json'

# Query data
curl --request POST --url http://localhost:8080/v1/instances/default/query --header 'Content-Type: application/json' \
  --data '{ "sql": "select * from ad_bids limit 10" }'

# Query explore API
curl --request POST --url http://localhost:8080/v1/instances/default/metrics-views/ad_bids_metrics/toplist/domain --header 'Content-Type: application/json' \
  --data '{
    "measure_names": ["measure_0"],
    "limit": 10,
    "sort": [{ "name": "measure_0", "ascending": false }]
}'

# Query profiling API
curl --request GET --url http://localhost:8080/v1/instances/default/null-count/ad_bids/publisher

# Get catalog info
curl --request GET --url http://localhost:8080/v1/instances/default/catalog

# Refresh source named "ad_bids_source"
curl --request POST --url http://localhost:8080/v1/instances/default/catalog/ad_bids_source/refresh

# Get available connectors
curl --request GET   --url http://localhost:8080/v1/connectors/meta

# List files in project
curl --request GET --url http://localhost:8080/v1/instances/default/files

# Fetch file in project
curl --request GET --url http://localhost:8080/v1/instances/default/files/-/models/ad_bids.sql

# Update file in project
curl --request POST --url http://localhost:8080/v1/instances/default/files/-/models/ad_bids.sql --header 'Content-Type: application/json' \
  --data '{ "blob": "select id, timestamp, publisher, domain, bid_price from ad_bids_source" }'

Documentation

Index

• type Options
• type Query
• type Runtime
  • func New(opts *Options, logger *zap.Logger) (*Runtime, error)
  • func (r *Runtime) Catalog(ctx context.Context, instanceID string) (*catalog.Service, error)
  • func (r *Runtime) CreateInstance(ctx context.Context, inst *drivers.Instance) error
  • func (r *Runtime) DeleteFile(ctx context.Context, instanceID string, path string) error
  • func (r *Runtime) DeleteInstance(ctx context.Context, instanceID string) error
  • func (r *Runtime) FindInstance(ctx context.Context, instanceID string) (*drivers.Instance, error)
  • func (r *Runtime) FindInstances(ctx context.Context) ([]*drivers.Instance, error)
  • func (r *Runtime) GetCatalogEntry(ctx context.Context, instanceID string, name string) (*drivers.CatalogEntry, error)
  • func (r *Runtime) GetFile(ctx context.Context, instanceID string, path string) (string, time.Time, error)
  • func (r *Runtime) ListCatalogEntries(ctx context.Context, instanceID string, t drivers.ObjectType) ([]*drivers.CatalogEntry, error)
  • func (r *Runtime) ListFiles(ctx context.Context, instanceID string, glob string) ([]string, error)
  • func (r *Runtime) OLAP(ctx context.Context, instanceID string) (drivers.OLAPStore, error)
  • func (r *Runtime) PutFile(ctx context.Context, instanceID string, path string, blob io.Reader, ...) error
  • func (r *Runtime) Query(ctx context.Context, instanceID string, query Query, priority int) error
  • func (r *Runtime) Reconcile(ctx context.Context, instanceID string, changedPaths []string, ...) (*catalog.ReconcileResult, error)
  • func (r *Runtime) RefreshSource(ctx context.Context, instanceID string, name string) error
  • func (r *Runtime) Registry() drivers.RegistryStore
  • func (r *Runtime) RenameFile(ctx context.Context, instanceID string, fromPath string, toPath string) error
  • func (r *Runtime) Repo(ctx context.Context, instanceID string) (drivers.RepoStore, error)
  • func (r *Runtime) SyncExistingTables(ctx context.Context, instanceID string) error

Types

type Options

type Options struct {
	ConnectionCacheSize int
	MetastoreDriver     string
	MetastoreDSN        string
	QueryCacheSize      int
}

type Query

type Query interface {
	// Key should return a cache key that uniquely identifies the query
	Key() string
	// Deps should return the source and model names that the query targets.
	// It's used to invalidate cached queries when the underlying data changes.
	Deps() []string
	// MarshalResult should return the query result for caching.
	// TODO: Also return estimated cost in bytes.
	MarshalResult() any
	// UnmarshalResult should populate a query with a cached result
	UnmarshalResult(v any) error
	// Resolve should execute the query against the instance's infra.
	Resolve(ctx context.Context, rt *Runtime, instanceID string, priority int) error
}

type Runtime

type Runtime struct {
	// contains filtered or unexported fields
}

func New

func New(opts *Options, logger *zap.Logger) (*Runtime, error)

func (*Runtime) Catalog

func (r *Runtime) Catalog(ctx context.Context, instanceID string) (*catalog.Service, error)

func (*Runtime) CreateInstance

func (r *Runtime) CreateInstance(ctx context.Context, inst *drivers.Instance) error

func (*Runtime) DeleteFile

func (r *Runtime) DeleteFile(ctx context.Context, instanceID string, path string) error

func (*Runtime) DeleteInstance

func (r *Runtime) DeleteInstance(ctx context.Context, instanceID string) error

func (*Runtime) FindInstance

func (r *Runtime) FindInstance(ctx context.Context, instanceID string) (*drivers.Instance, error)

func (*Runtime) FindInstances

func (r *Runtime) FindInstances(ctx context.Context) ([]*drivers.Instance, error)

func (*Runtime) GetCatalogEntry

func (r *Runtime) GetCatalogEntry(ctx context.Context, instanceID string, name string) (*drivers.CatalogEntry, error)

func (*Runtime) GetFile

func (r *Runtime) GetFile(ctx context.Context, instanceID string, path string) (string, time.Time, error)

func (*Runtime) ListCatalogEntries

func (r *Runtime) ListCatalogEntries(ctx context.Context, instanceID string, t drivers.ObjectType) ([]*drivers.CatalogEntry, error)

func (*Runtime) ListFiles

func (r *Runtime) ListFiles(ctx context.Context, instanceID string, glob string) ([]string, error)

func (*Runtime) OLAP

func (r *Runtime) OLAP(ctx context.Context, instanceID string) (drivers.OLAPStore, error)

func (*Runtime) PutFile

func (r *Runtime) PutFile(ctx context.Context, instanceID string, path string, blob io.Reader, create bool, createOnly bool) error

func (*Runtime) Query

func (r *Runtime) Query(ctx context.Context, instanceID string, query Query, priority int) error

func (*Runtime) Reconcile

func (r *Runtime) Reconcile(ctx context.Context, instanceID string, changedPaths []string, forcedPaths []string, dry bool, strict bool) (*catalog.ReconcileResult, error)

func (*Runtime) RefreshSource

func (r *Runtime) RefreshSource(ctx context.Context, instanceID string, name string) error

func (*Runtime) Registry

func (r *Runtime) Registry() drivers.RegistryStore

func (*Runtime) RenameFile

func (r *Runtime) RenameFile(ctx context.Context, instanceID string, fromPath string, toPath string) error

func (*Runtime) Repo

func (r *Runtime) Repo(ctx context.Context, instanceID string) (drivers.RepoStore, error)

func (*Runtime) SyncExistingTables

func (r *Runtime) SyncExistingTables(ctx context.Context, instanceID string) error