[Wranglers] Databases that added support for Translytical workloads in 2024

[Ali, Saqib]

A nice overview of how the GaussDB supports HTAP workloads. Unlike SQL DB
in Microsoft Fabric that provides two different end-points, one for OLAP
and one for OLTP loads separately, GaussDB offers a single end-point that
can be used for both Transactional and Analytical queries. GaussDB makes
the decision behind the scene to route the query to Row Storage (OLTP) or
to the Columnar Storage (OLAP). This decision is based on both query type
and the query cost:
https://www.youtube.com/watch?v=x2tMHqKnCSQ



On Sun, Jan 12, 2025 at 5:55 AM Ali, Saqib <docbook.xml at gmail.com> wrote:

> One thing I like about the hybrid tables in Snowflake is that it supports
> all the advanced analytics functions like match_recognize, dynamic pivot
> etcs.
>
> On Fri, Jan 10, 2025 at 5:55 PM Ali, Saqib <docbook.xml at gmail.com> wrote:
>
>> Forgot to mention that Snowflake also has Hybrid tables that support OLTP
>> workloads with ability to define and enforce Primary Keys and Unique Key
>> constraints.
>>
>> In general, I think, analytical databases will want to be close to
>> operational/transactional data to enable insight from operational data as
>> easily as analytical databases. We will see advanced analytical SQL
>> functions being supported in these translytical / HTAP / Hybrid tables.
>>
>> On Fri, Jan 10, 2025 at 4:47 PM Ali, Saqib <docbook.xml at gmail.com> wrote:
>>
>>> Here are three databases that added support for translytical workloads
>>> in 2024:
>>>
>>>
>>>    1. SQL DB in Microsoft Fabric. All data is replicated in near-real
>>>    time to open-source delta parquet format in OneLake. When you deploy a SQL
>>>    database in Fabric, you get two endpoints: an endpoint to the database and
>>>    a “SQL analytics endpoint” that allows querying the replicated data in
>>>    OneLake. This unifies operational and analytics work without having to move
>>>    any data. With the SQL analytics endpoint, you can query using the data
>>>    warehouse engine.
>>>    2. TiDB. Real-time replication of data from OLTP Table to Columnar
>>>    TiFlash tables. Analytical queries are served by the TiFlash Columnar
>>>    tables.
>>>    3. AlloydDB. Enable the Columnar Engine on the AlloyDB and then
>>>    either manually add columns to the Columnar Engine or use
>>>    auto-columnarization, which analyzes your workload and automatically adds
>>>    columns.
>>>
>>>
>>> Saqib
>>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://analyticsengineering.net/pipermail/wranglers/attachments/20250527/05e2ee6d/attachment.htm>