MongoDB .local Munich 2019: MongoDB Atlas Auto-Scaling
0 likes380 views
The document discusses MongoDB's Atlas autoscaling, which automatically adjusts computing resources based on server load to optimize performance and cost. It outlines the mechanism of autoscaling, user notifications, metrics for scaling decisions, and scenarios when it is beneficial or not. Additionally, it highlights future enhancements and encourages users to engage with resources and support available from MongoDB.
![#MDBLocal @mlynn
{
_id: ObjectID(”A14GDB23163…”),
Name: ”Michael Lynn”,
Title: ”Director, Principal DevRel”,
Company: ”MongoDB”,
Location: ”Philadelphia, PA”,
Twitter: ”@mlynn”,
Previous: {
Titles: [”Engineer”,”Architect”]
},
Programs: [“MongoDB for Startups”,
“MongoDB for Schools”]
}
Who is this guy?](https://image.slidesharecdn.com/400pmmongodbatlas-autoscaling-191017151813/85/MongoDB-local-Munich-2019-MongoDB-Atlas-Auto-Scaling-3-320.jpg)
MongoDB .local Munich 2019: MongoDB Atlas Auto-Scaling
- 1. Atlas Autoscaling Michael Lynn Director, Principal Developer Advocate
- 2. Agenda What is it? How does it work? When to use it? What’s Next? Atlas Autoscaling
- 3. #MDBLocal @mlynn { _id: ObjectID(”A14GDB23163…”), Name: ”Michael Lynn”, Title: ”Director, Principal DevRel”, Company: ”MongoDB”, Location: ”Philadelphia, PA”, Twitter: ”@mlynn”, Previous: { Titles: [”Engineer”,”Architect”] }, Programs: [“MongoDB for Startups”, “MongoDB for Schools”] } Who is this guy?
- 4. #MDBLocal @mlynn ● Access to MongoDB Platform and Tools ● Technical Support and Advising ● Network and Community Access ● Marketing Support ● Access to MongoDB University ● Access to MongoDB Atlas ● Curriculum Engineering Support ● Network and Community ● Marketing Support Interested? Reach out, DM’s welcome: @mlynn [email protected]
- 5. #MDBLocal @mlynn TIME ACTIVITY LAUNCH RELEASE / UPDATE NEWS / MARKETING DEVELOPMENT RELEASE / UPDATE In an ideal world...
- 6. #MDBLocal @mlynn TIME ACTIVITY LAUNCH RELEASE / UPDATE SUN SPOTS NEWS / MARKETING DEVELOPMENT In reality...
- 8. #MDBLocal @mlynn At some point, we need to think about deployment...
- 9. #MDBLocal @mlynn Enter our hero...
- 10. #MDBLocal @mlynn TIME ACTIVITY M0 Tier Cluster Capacity Utilization
- 11. #MDBLocal @mlynn TIME ACTIVITY M0 Tier Cluster Capacity Utilization
- 12. #MDBLocal @mlynn TIME ACTIVITY M0 Tier Cluster Capacity Utilization M100 Tier Cluster Excess Capacity
- 13. #MDBLocal @mlynn TIME ACTIVITY M0 Tier Cluster Capacity Utilization M100 Tier Cluster CapacityM40 Tier Cluster Excess Capacity Capacity Over Capacity
- 14. #MDBLocal @mlynn TIME ACTIVITY LAUNCH RELEASE / UPDATE SUN SPOTS NEWS / MARKETING DEVELOPMENT Planning for changes in utilization... Yep - planned for that Yep - planned for that Viral Moment… Uhh… didn’t see that coming
- 15. #MDBLocal @mlynn Free Tier Upgrade to M10 Upgrade to M30 Downgrade to M30 Upgrade to M40
- 16. #MDBLocal @mlynn CPU Utilization Activity Response Time Memory
- 17. #MDBLocal @mlynn MongoDB Atlas Metrics / Monitoring
- 18. #MDBLocal @mlynn Setting Alerts ● Asserts ● Opcounters ● Memory ● Connections ● Queues ● Page Faults ● Cursors ● Network ● Replication Oplog ● DB Storage ● Wired Tiger - Tickets ● System, Disk
- 19. So now you’re notified… now what?
- 20. #MDBLocal @mlynn
- 21. #MDBLocal @mlynn What is Auto-Scaling? a method used in cloud computing, whereby the amount of computational resources in a server farm, typically measured in terms of the number of active servers, scales automatically based on the load on the farm.
- 23. #MDBLocal @mlynn
- 24. #MDBLocal @mlynn What do we do? Autoscaling Compute Storage ● Vertical scaling between Atlas Tiers (ex: M20 to M30) ● Only for tiers with standard CPU and Disk ● Can scale both up and down ● Up - keep up with performance ● Down - save $$s ● Scale storage capacity within a tier ● Automatically keeps up with write workloads
- 25. #MDBLocal @mlynn How it works? User Atlas Metrics Atlas predicts future utilization Executes autoscaling decisions Check user specified conditions Automated hardware provisioning Atlas automatically changes cluster’s computation capacity with no downtime Utilization metrics are captured in real time User Alerts to users
- 26. #MDBLocal @mlynn How does it work? CPU Util > 75% OR Memory > 75% UP For at least 1 hour
- 27. #MDBLocal @mlynn How does it work? CPU Util < 50% AND Memory < 50% DOWN For at least 72 hours
- 28. #MDBLocal @mlynn When to use Autoscaling?
- 29. #MDBLocal @mlynn Don’t care about time?
- 30. #MDBLocal @mlynn Don’t care about Money?
- 31. #MDBLocal @mlynn Don’t care about Quality?
- 32. #MDBLocal @mlynn Impact so Far % of Atlas Clusters which we would have downgraded/upgraded $ figures of how much money we would save customers
- 34. #MDBLocal @mlynn Quick UI Walkthrough
- 35. #MDBLocal @mlynn When not to use auto-scaling CHOPPY SPIKEY
- 37. #MDBLocal @mlynn The future… Increased reaction speed.
- 38. #MDBLocal @mlynn The future… Incorporating Seasonality
- 39. #MDBLocal @mlynn The future… Incorporating Core Product Features Indexing Sharding
- 41. #MDBLocal @mlynn ● Access to MongoDB Platform and Tools ● Technical Support and Advising ● Network and Community Access ● Marketing Support ● Access to MongoDB University ● Access to MongoDB Atlas ● Curriculum Engineering Support ● Network and Community ● Marketing Support Interested? Reach out, DM’s welcome: @mlynn [email protected]
- 42. #MDBLocal @mlynn Suggestions ● Start Small ● Build, Test, Iterate ● Plan for Scale in all Directions
- 43. #MDBLocal @mlynn