Presented by Achille Brighton, Principal Consulting Engineer, MongoDB
Experience level: Introductory
New to MongoDB? We'll provide an overview of installation, high availability through replication, scale out through sharding, and options for monitoring and backup. No prior knowledge of MongoDB is assumed. This session will jumpstart your knowledge of MongoDB operations, providing you with context for the rest of the day's content.
10. 10
MongoDB in a Nutshell
• A document store: more complex data than relational
• Query language is just about as expressive as SQL
• Much greater focus on scalability
– replication (high availability)
– sharding (volume and throughput)
12. 12
DBA Tasks for MongoDB
• There are no schemas
• Indexes are more important than in relational DBs
• Responsibility for security
• Likely dealing with a lot of servers
– monitoring
– backup
– deployment
14. 14
MongoDB Cloud Manager
• Comprehensive, web-based management solution
– modules for monitoring, backup, automation
• Cloud Manager
– we run it for you
– enhances interaction with MongoDB tech support
• OpsManager (on prem)
– you run it yourself
– all under your control, nothing leaves your network
16. 16
Backup
• Registers itself as a hidden replica,
works cluster-wide
• Snapshot every six hours,
plus oplog, thus
point-in-time restore
• Requires 3-4x original storage for typical retention policy
• Can do 100s of GB/day over the net
• On-prem version available for enterprise customers
17. 17
Automation
• Deploy, configure and control
MongoDB on your own hardware
• Create complex configurations
with dozens, hundreds of servers
in seconds
• Orchestrate complex tasks such as version upgrades,
change of storage engine
• Each server runs an autonomous automation agent
18. 18
Summary
• Paradigm shift in databases is happening
• MongoDB is extremely attractive to modern application
development: complex data structures, huge volume
• For DBAs, some things are different, some things stay
the same: no schemas, but many servers
• Know as much as you can!
Exciting
DB Engines, popularity league
If we extrapolate the growth, we hope we’ll reach Eclipse in 2019
Sentiment in market.
This reflects sentiment in the market
NoSQL diff is bigg
Json consists of key value paiers.
Data types
Address is an array Age is a number
Closer to app
Dynamic Schema, not a problem
Find subset
You can find a document based ono subset
You can do very complicated things.
The advantage it does not look like SQL, but it can do almost as much.
Let’s look at ways you can strucutre the data
All the operations go to the primaries, we have three databases.
Either scale horizontally or vertically
Designed for scale
Horizontally
Election built right from the start
Analogy so let’s say that each one of the nodes has only so much capacity,
You can get more capacity. What about reliability.
We have customers with thousands of shard. And you can’t lose any, otherwise you lose data.
Didn't invent.
You can’t lose shards
You don’t have
Licensing cost is way lower
15 servers
It’s a single unix process
You’re responsible for security.
Flight traffic.
Flights over europe
Rolls royce HQ in london gets a live feed from every engine.
If there’s an issue: ping the pilot to look outside and see