Ultipa Graph approaches version 5.0

Written By:
Content Copyright © 2024 Bloor. All Rights Reserved.
Also posted on: Bloor blogs

Ultipa Graph approaches version 5.0 banner

Ultipa Graph, one of the newer entrants to the graph database space, is approaching its 5.0 version update, with the beta due to hit later this month. As one of the graph’s main selling points is its performance, it should come as no surprise that it has been the focus of this update.

For example, Ultipa is already able to scale horizontally in a way that some of its contemporaries struggle to, using a cluster-based design supported by distributed consensus and similar techniques to create a distributed graph architecture. But the rub is that scaling out like this currently requires a DBA to manually configure each cluster, making the configuration process considerably slower and more painful than it ought to be and hamstringing any performance gains before they can be fully realised. That is, until 5.0, which promises to automate this process, making the performance enhancements it offers practical, not just theoretical.

Ultipa 5.0 is also set to implement Pregel, Google’s framework for large-scale graph processing. This is used, for instance, to power distributed graph algorithms. In addition, the update will feature a redesign to the product’s storage and compute engine, principally to better enable high density computing (HDC). It will also position storage and compute locations as close together as possible, to minimise the need for data migration, and it will enable the lifting-and-shifting of data in order to handle n-hop queries (for some reasonably large n – 10, say) in a faster timeframe. We are told that the performance of data ingestion has also been improved, especially for large loads.

Moreover, the updated Ultipa will be fully compatible with GQL (due to be officially published in late April or May – assuming there are no holdups, which is not a given) and backwards compatible with 4.x versions of the product. It will also be able to work with large language models (LLMs) to help prevent hallucinations (by, for example, conducting causality searches – notably, on a white box basis).

In short, Ultipa version 5.0 represents a meaningful iteration of the product’s existing design and its accompanying strengths. It isn’t adding any new strings to the product’s bow, but that’s no bad thing: Ultipa is, and will remain, a highly performant graph database.