Corporate data platforms often consist of multiple data warehouses and data lakes for batch processing and ad-hoc analysis, in addition to a variety of upstream and downstream components. Thus, they require a big team of maintainers who must have a good command of all tools and tech stacks.

Within a data platform, parallel processing systems can cause redundant data storage in multiple components. Moreover, as data varies in access frequency, the strong coupling of hot and cold data is a big waste of storage resources.

For companies, system scalability is often prioritized over single-node performance. However, long and complicated data pipelines result in high latency in data processing, which is the opposite of what online reporting and data analytics need.

The complexity of component integration makes it hard to introduce new functionality to the existing data platform. This undermines a company's adaptability to changes and new business needs. Meanwhile, the gap between cloud and on-premise deployments is another barrier to system openness.