Issue #266: NVIDIA unveils a direct path between storage and GPUs for data science workloads

#266 — August 9, 2019

Life After Hadoop: How NVIDIA Wants to Accelerate Data Science — We’ve featured NVIDIA’s data science ambitions before and now here’s an article from two NVIDIA engineers outlining how NVIDIA sees GPU-based tooling taking things to the next level, including GPUDirect Storage, a new way to directly connect storage and GPUs.

Josh Patterson

Building Petabyte-Scale Analytics with BigQuery and HyperLogLog — Permutive provides a data and analytics platform for publishers and BigQuery and HyperLogLog play a big role in keeping things fast and accurate.

Paulius Imbrasas

eBook: Best Practices for Optimizing Postgres Query Performance — Learn how to get a 3x performance improvement on your Postgres database and 500x reduced data loaded from disk in this free pganalyze eBook.

pganalyze sponsor

Comparing Postgres's JSONB with Couchbase — Can Postgres really take the place of a ‘NoSQL’ document-oriented database? It’s interesting to see something like this from a document-oriented database vendor, especially as it’s reasonably fair to Postgres (while still focusing on Couchbase’s strengths, naturally).

Denis Rosa (Couchbase)

Build Highly Available MySQL Apps using Amazon Aurora Multi-Master — The newly available Amazon Aurora Multi-Master brings further high availability features to Aurora, Amazon’s MySQL compatible database service.

Amazon Web Services

Quick bytes:

PostgreSQL 11.5, 10.10, 9.6.15 and other minor releases have been released to fix two security issues and 'over 40 bugs.'
The first ever Distributed SQL Summit is taking place in San Jose on September 20.
From October 1, 2019, Microsoft's licensing terms will impact users using on-prem services on 'dedicated hosted cloud services' (e.g. AWS).
HPE has acquired MapR, the Hadoop company that has been facing an uncertain future recently.
BigQuery now has support for persistent user defined functions.
In anticipation of PostgreSQL 12 adding support for SQL/JSON, Markus Winand has been testing and comparing the SQL/JSON functionality of other databases including SQLite, MariaDB, MySQL, and SQL Server. Chart here.
BlazingSQL, a RAPIDS-based SQL engine, is now open source. Another interesting part of the NVIDIA data science story.

💻 Jobs

Lead Cloud Security Engineer — Make an impact as our first Security Engineer, focusing on driving innovation & best practices around our cloud security efforts.

Cockroach Labs

DB Developers Are in Demand on Vettery — Ready for a bold career move? Make a free profile, name your salary, and connect with hiring managers from top employers today.

Vettery

📒 Everything else

How I Destroyed Multiple Production Databases — A sysadmin ‘horror story’ from the 90s taught this engineer some lessons for the future.

Jan Gerrit Kootstra

How Stack Overflow Does App Caching with Redis — An interesting, technical writeup of how popular developer Q&A site Stack Overflow uses Redis to keep things running smoothly.

Nick Craver (Stack Overflow)

The DevOps Perspective on Database Coexistence — Whether you're in DevOps, development, or database administration, download the full SQL Migration whitepaper for free, from Studio 3T.

Studio 3T sponsor

MariaDB vs MySQL, A Database Technologies Rundown — If you’ve just thought of MariaDB as a sort of MySQL spinoff, there’s a bit more to it than that..

Tonino Jankov

You Don't Need an ER Diagram to Understand Your Database — “ERDs are great for visualizing databases but are not suitable as the documentation in itself”

Piotr Kononow

MontyDB: A Tiny Clone of MongoDB in Python — An interesting bit of fun here for Python developers. It’s a TinyDB-inspired, pure Python MongoDB-a-like!

David Lai

LF: A Fully Decentralized, Fully Replicated Key/Value Store — It’s billed as ‘beta’ software, but this is an interesting experiment in creating a fast data store for ‘small but critical’ pieces of information (think etcd but suitable for open, decentralized systems).

ZeroTier, Inc.