Skip to main content

Zerobus Ingest connector limitations

Preview

The Zerobus Ingest connector is in Public Preview. To try it, contact your Databricks account representative.

This page lists limitations when using the Zerobus Ingest connector in Lakeflow Connect.

Durability

We only support single availability zone (single-az) durability. This means our service may experience downtime if our zone is unavailable.

Throughput restrictions

  • Same AZ
  • 100MB/second per stream (benchmarked with 1KB-sized messages)
  • 15,000 rows per second per stream

Maximum throughput can be achieved when a client app and endpoint are in the same geo region. If you need higher throughput, contact your Databricks account representative.

Delivery guarantees

The connector only provides at-least-once guarantees.

Workspace

The workspace must be in one of the following regions:

  • us-east-1
  • us-east-2
  • us-west-2
  • eu-central-1
  • ap-southeast-1
  • ap-southeast-2
  • ap-northeast-1
  • ca-central-1
  • eu-west-1

Target table

  • We support writing to managed Delta tables only. We do not support writing to managed storage.

  • We do not support recreating a target table.

  • We only support table names with ASCII letters, digits, and underscores.

  • The following regions are supported:

    • us-east-1
    • us-east-2
    • us-west-2
    • eu-central-1
    • ap-southeast-1
    • ap-southeast-2
    • ap-northeast-1
    • ca-central-1
    • eu-west-1

Schema evolution

We do not support schema evolutions on the target table.

Protobuf schema

  • The protobuf schema definition must match 1:1 with the Delta table schema. If not, the API returns an error. This includes:

    • Number of columns
    • Column names
    • Column optionality must match between Delta and Protobuf schemas
  • We do not support proto schemas with more than 2000 columns.

  • We only support table and column names with ASCII letters/digits/underscores.

  • We do not support using a different proto schema for “stream creation” and “ingest record”.

Record size

Each individual message is limited to 4MB in size. Record size max is 4,194,285 Bytes. Headers required for communication take up 19 Bytes.

Type support

Delta types

Protobuf types

Integer

Int32

String

String

Float

Float

Long

Int64

Short

Int32

Double

Double

Boolean

Bool

Binary

Bytes

Date*

Int32

Timestamp**

Int64

Array<TYPE>

repeated TYPE

Map<K,V>

map<K,V>

Struct<FIELDS>

message Nested { FIELDS }

*Should be converted to int32 (number of days since epoch)

** Should be converted to int64 (epoch time in microseconds)

***The map Protobuf syntactic sugar is available only for Protobuf compilers version 3 and above