Recently the Hive hosted a panel discussion on the pros and cons of various approaches to in-memory databases. Yours truly served as moderator/referee as Oracle, MemSQL, IBM, Aerospike, and Pivotal sent their biggest brains. Check out who was on our eminent panel here.
Each was able to represent his thoughts on a wide range of questions, including:
- What are the advantages of an IMDB? Is it just speed or anything else?
- What applications are best suited or most common for IMDBs? What aren’t?
- Why now? Is it just cost of memory or some other factors?
- ACID compliance is important for a reason—what about durability for IMDBs?
- Are the boundaries between memory and storage blurring?
- How does cloud change things again in terms of resource provisioning?
- What is the biggest advantage of your particular approach? Where doesn't it work well?
Some of the more humorous quotes of the evening:
- "Anyone come here thinking IMDB was a way a to look up what actors were in an old movie?"
- "See the Mayweather Pacquiao fight? Well that ring had only two people to referee, I have five."
- "Anyone here looking to build a 50 gig database?" "Ok, then."
- "Our platform supports every workload..." "Get out! Out!"
- "I know there are still more questions, but I can say with confidence that these vendors would be happy to talk with you more at any time."