January 2014 - Toronto MongoDB Meetup

Hosted by Kevin Cearns and Nick Boccone at Bento Miso. Paul Oude-Reimerink and Chad Tindel from MongoDB Inc.

Walked through resources like Mongo University and Mongo Monitoring service, and then had a beginner-intermediate level QA session with the TenGen guys, talking about use cases, experiences as solutions architects, and some technical stuff. The meetup lasted about 2 hours, then we all walked out of the building like penguins because it was particularly icy that night.

I liked seeing the variety of skill levels in one room, and hearing from people who had experienced exactly the struggles I have.

Notes

• Mongo University has free online courses, certifications, videos are always on YouTube when courses aren’t running

• Mongo Monitoring Service has some free monitoring services.

• There is a Canadian MongoDB Sales guy – Paul

• Walmart Canada briefly had a site that used Mongo because requirements couldn’t be nailed to a schema.

• Multiple replica sets running on the same hardware is useless. To ensure each set is on different hardware in the cloud, put them in different AWS regions or availability zones or hosting providers

• Applications commonly to mix SQL and nosql databases because each has its own strengths for storing/accessing data

• A giant insurance company combined data from 70 different dB’s in different departments into a MongoDB so customer service reps could get a complete view of one customer’s info in one place. Way more effective than a relational dB.

• No transactions coming. Is hard to scale!

• Text Index and fulltext searching coming in some v2.6, in QA now. Maybe February.

• People doing full text search now often copy data to another DB that is searchable.

• Mongo sales people spend lots of time telling people not to think RBDMS style.

• Write throughput of 10,000s per second is not unreasonable on some unknown hardware.

• There is a tool for checking the size of your working set.

• SSDs are great for write heavy applications. Eg. streaming lots of tweets to a DB.

• Presharding speeds up massive dumps of data into sharded db environments.

• 85% of people aren’t using sharding. Most do use replica sets.

• Shard key values can’t be edited.

• Freenode IRC channel is helpful.