Reflections on Learning GBase 8a

By Liu Kai

“After a long and fruitless search, you find it effortlessly.”

I had been working on traditional relational database projects, such as Oracle, and never had a chance to explore columnar databases. To my surprise, while working at a bank in Chongqing, I got the opportunity to systematically learn GBase 8a — a pleasant surprise.

When I registered, I made a blunder: I didn’t join the May class group in time. By coincidence I ended up in the June class. Special thanks to Teacher Shen, who helped me switch to the May class, giving me the chance to learn with so many like-minded peers.

“The road is long and full of obstacles, yet if you keep going, you will reach the destination.”

“The road is long and full of obstacles, yet if you keep going, you will reach the destination” — this truly reflects my feelings when looking at GBase’s release history. From its debut in 2010 to the current GBase 8a and 8a Cluster MPP clusters, it has traversed a long and painstaking journey, each iteration truly eye-catching. Its unique columnar storage, transparent adaptive compression, smart indexes, two-way parallelism, high-speed data loading, automatic memory management, and semantic optimization showcase GBase's “three highs” advantages over traditional databases: high performance, high cost-effectiveness, and high availability. It is said that it can reduce I/O to 1/10 of that of traditional databases, filling me with anticipation.

In projects I worked on previously, I encountered two pain points: an insurance company in the capital, Beijing, used Oracle for its reporting database, with over 60 million records per period, employing period & primary key for list-range partitioning. Queries on this table were fine for the first two years, but when total records exceeded one billion, the database would freeze, despite having fairly decent hardware. Another insurance company in Shenzhen wanted to use reports for ad-hoc analysis on a large Oracle table, but the data simply couldn’t be pulled out. In recent years, BI has advanced rapidly, with mature and easy-to-use tools emerging domestically and internationally like waves. Yet, I didn’t expect that the database would become the weakest link — it was hard for me to accept. In the wave of “de-IOE” (moving away from IBM, Oracle, and EMC), GBase from GBASE (General Data Technology) emerged abruptly, a fully homegrown product, like a glimmer of light in the darkness, heralding the coming dawn. This truly excited me, yet also made me quite anxious, just like about to meet a long-time crush — afraid she won’t show up, and afraid she will but won’t be exactly as I imagined.

“A ‘Slap’ and a ‘Sweet Date’”

Let’s talk about this training. It’s a fifteen-day course, requiring only 1–2 hours per day. From GBase’s history and features to deployment, installation, regular use, and monitoring — it is very comprehensive. You have to pass three quizzes before you can take the final certification exam; those who perform well and actively participate may get exam fees and certificate fees waived. I must say, the arrangement is excellent: you get both the “sweet date” and the “slap,” haha. I like this model — it sparks interest, encourages broad participation, and avoids the embarrassment of the training becoming a mere formality. I highly recommend those who haven’t joined yet to quickly sign up for the June class. I’ve also forwarded the registration link to my team and strongly urged everyone to participate if they have time, to broaden their knowledge and learn something truly useful.