Qlik Replicate

Overview

Qlik Replicate is universal data replication software, built to move data between databases, warehouses and streaming targets in real time, with a graphical console rather than code.

Capture is log-based and marketed as agentless: it scans transaction logs remotely, so nothing is installed on the production source, with mainframe and SAP extraction the exceptions.

Tasks have two phases, a full load of the selected tables followed by change processing, with committed changes grouped by transaction and sorted chronologically before being applied.

Know its lineage, because it explains the product. This is Attunity Replicate; Qlik acquired Attunity in 2019 for around $560m, and the help site still carries Attunity-era release notes.

It is the client-managed half of Qlik's portfolio. The cloud-managed sibling is Qlik Talend Cloud, which is a different product rather than a hosted Replicate, so there is no managed Replicate service.

Its real strength against the open-source option is reach. Sources include IBM IMS and VSAM on the mainframe, Db2 for z/OS and iSeries, SAP applications and extractors, Salesforce, Teradata, SAP HANA and Sybase ASE.

Targets are genuinely warehouse-native, covering Snowflake, Redshift, BigQuery, Synapse, Microsoft Fabric, Databricks Delta and Cloudera Iceberg, alongside Kafka, Kinesis, Pub/Sub and Event Hubs.

It is replication only. There is no streaming SQL engine, no windowing and no in-flight enrichment; Qlik puts transformation in Compose and Talend instead.

Read the Kafka guarantees carefully. Directing everything to one topic and partition guarantees ordered delivery of transactions and of changes within them, but across partitions the consumer has to buffer and re-sort, and Qlik publishes no at-least-once or exactly-once claim.

On schema changes it propagates ALTER TABLE automatically, but with a documented catch: adding or modifying a column with a default does not replicate the default, and the column arrives as null.

Key features and capabilities

The same headings are used for every change data capture & replication entry, so two tools can be read side by side.

How changes are captured

• Log-based change capture, scanning transaction logs remotely with no agent on the production source
• Two phases per task, a full load of selected tables then change processing
• Committed changes are grouped by transaction and sorted chronologically before being applied
• SQL Server has two source endpoints, the log reader and a Microsoft CDC endpoint, the latter the only route to Azure SQL Database
• Incremental snapshot is not published; re-reading means a per-table full load

Source databases

• Oracle 19c to 26ai, SQL Server 2016 to 2025, PostgreSQL 14 to 18, MySQL 8.4, MariaDB and Percona
• Mainframe through IBM IMS and VSAM, plus Db2 for z/OS, iSeries and LUW, and Informix
• Enterprise sources others lack, including SAP HANA, Sybase ASE, Teradata and MongoDB
• SAP application, SAP Extractor, SAP ODP, SAP OData and Salesforce
• Generic ODBC sources carry an explicit warning not to use them without Professional Services approval

Targets

• Streaming to Kafka, Confluent Cloud, Kinesis, Google Pub/Sub and Azure Event Hubs
• Warehouses including Snowflake, Redshift, BigQuery, Azure Synapse, Microsoft Fabric and Teradata
• Lakehouse through Databricks Delta and Cloudera Iceberg
• Object storage and Hadoop, including S3, Google Cloud Storage and Cloudera
• A generic Iceberg writer and Apache Hudi are not published

Delivery and consistency guarantees

• Directing all messages to one topic and partition guarantees ordered delivery of transactions and their changes
• Across multiple partitions the consumer must buffer and re-sort by change sequence and transaction id
• Every message carries a transaction id and a growing change sequence, so boundaries are recoverable
• No at-least-once or exactly-once claim is published for the Kafka target
• Qlik warns that transactional mode against endpoints without constraints can produce duplicate target rows

Handling schema changes

• ALTER TABLE is propagated automatically by comparing previous and new table metadata
• The supported DDL set varies by target, covering truncate, drop, create, add, rename and convert
• A DDL history control table records schema events
• Adding or modifying a column with a default does not replicate the default, and the column arrives null
• Renaming a source table while a task is stopped, and reallocating primary key columns, are unsupported

Monitoring and recovery

• The console monitors full load and change processing per task, with a notifications list and message centre
• Notifications can be sent to the Windows event log and to named email recipients
• Qlik Enterprise Manager gives central monitoring across servers, with its own notification rules
• Enterprise Manager can run in a Windows high-availability cluster, licensed to the cluster name
• Tasks resume from their stored position, with individual tables reloadable

Where it runs and what it costs

• Self-hosted only, on Windows or Linux, with Docker supported since version 6.4
• No Kubernetes operator is published
• Cloud marketplace listings exist but are client-managed on your own instances
• There is no managed Replicate service; Qlik Talend Cloud is a separate product
• A unit is one single-core relational source connected to one target

Pricing

Price on request$10,000 a year for one source and target

Effectively quote-only, with one published figure. Qlik's own pricing page covers only Qlik Cloud Analytics and does not mention Replicate at all. Its AWS Marketplace listing prices a 12-month contract at $10,000, where one unit covers a single-core relational source and one target, with the software client-managed so infrastructure is extra. Multi-core licences and additional sources or targets are available only through a private offer, and listings differ by selling entity, so anything realistic is negotiated through sales.

Vendor pricing page →

Demos and videos

• Qlik Replicate explainer and demo Product demo · YouTube
• Qlik Replicate, a hello world demonstration Tutorial · YouTube

About Qlik

Qlik was founded in 1993 in Lund, Sweden and is now headquartered in King of Prussia, Pennsylvania; its own pages do not name its founders. It is private: Thoma Bravo took it private in 2016 and remains majority owner, and in a deal signed in November 2024 and closed in 2025 sold a significant minority stake to a subsidiary of the Abu Dhabi Investment Authority, on undisclosed terms. Saugata Saha became chief executive on 31 July 2026, succeeding Mike Capone. Qlik acquired Attunity, the origin of Replicate, in May 2019 for about $560m, and Talend in 2023. It claims over 40,000 customers.

Founded 1993 · King of Prussia, Pennsylvania · qlik.com