Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

MongoDB Days Silicon Valley: MongoDB Support Cases: The Blockbusters

612 views

Published on

Presented by Daniel Coupal, Technical Services Engineer, MongoDB

Experience Level: Introductory

In his two years as a Technical Support Engineer, Daniel has worked with hundreds of MongoDB customers. Daniel will take you on a ride through some of his most memorable customer stories, told through the plots of your favorite Hollywood films. Get to hear about some of more unusual MongoDB use cases, the idiosyncratic choices behind them, and their path to success. You'll laugh, you'll cry, and you'll save yourself from making the common mistakes.

Published in: Technology
  • Be the first to comment

  • Be the first to like this

MongoDB Days Silicon Valley: MongoDB Support Cases: The Blockbusters

  1. 1. MongoDB Support Cases: The Blockbusters Daniel Coupal Technical Services Engineer Palo Alto
  2. 2. 3 Support Team
  3. 3. 4 Goal of the Presentation •  Remember 3 things from this presentation •  Make 1 change in your application or deployment •  Good return on your hour spent in this presentation
  4. 4. 5 #1 - “Toto, I've a feeling we're not in Kansas anymore”
  5. 5. 6 #1 - “Toto, I've a feeling we're not in Kansas anymore” ü The Wizard of OZ (1939)
  6. 6. 7 #1 – The NoSQL World is different
  7. 7. 8 #1 – The NoSQL World is different Relational World NoSQL World 1 - Model the data 1 - Define the access to the data 2 - Write the queries to access the data 2 - Model the data
  8. 8. 9 #2 - “Houston, we have a problem”
  9. 9. 10 #2 - “Houston, we have a problem” ü Apollo 13 (1995)
  10. 10. 11 #2 – Have a Test Environment •  Best –  identical cluster –  restore production data in test/staging/QA environment •  Acceptable –  less shards with identical hardware •  Less desirable –  smaller hardware –  non representative data
  11. 11. 12 #3 – “Elementary, my dear Watson”
  12. 12. 13 #3 – “Elementary, my dear Watson” ü The Adventures of Sherlock Holmes (1905, 1921, 1939, 1984)
  13. 13. 14 #3 – Look for Clues of Bad Queries (slow queries in mongod log) •  2.6 format •  2015-09-23T07:06:15.465-0400 [conn21987] query mydb.coll1 query: { tr: { $lte: "2014-08-24-day" }, a: "", b: "min" } planSummary: COLLSCAN ntoreturn:0 ntoskip:0 nscanned:13798085 nscannedObjects:13798085 keyUpdates:0 numYields:267371 locks(micros) r:105591398 nreturned:0 reslen:20 812656ms •  3.0 format •  2015-10-15T19:48:15.095+0000 I QUERY [conn1844] query mydb.coll2 query: { $readPreference: { mode: "primary" }, $query: {key: "ZrzfmfFlUV3i8iARTyXWD0Qxa8I=" } } planSummary: COLLSCAN ntoskip:0 nscanned:0 nscannedObjects:83256 keyUpdates:0 writeConflicts:0 numYields:650 nreturned:0 reslen:20 locks:{ Global: { acquireCount: { r: 1302 } }, Database: { acquireCount: { r: 651 } }, Collection: { acquireCount: { r: 651 } } } 121ms
  14. 14. 15 #3 – Look for Clues of Bad Queries •  mtools for plots and stats on queries
  15. 15. 16 #3 – Look for Clues of Bad Queries •  CloudManager –  Suggested indexes
  16. 16. 17 #4 - “Round up the usual suspects”
  17. 17. 18 #4 - “Round up the usual suspects” ü Casablanca (1942)
  18. 18. 19 #4 – Settings in Production Notes •  Hardware –  Using NAS –  Using other FS than xfs on Linux for Wired Tiger –  Using other FS than ext4 or xfs on Linux for MMapV1 –  Maxed out spinning disks •  OS Settings –  ulimits too low –  too high readahead for disks –  Transparent Huge Pages on –  NUMA on
  19. 19. 20 #4 – Settings in Production Notes •  Tools –  Production notes –  Mdiag (in Github)
  20. 20. 21 #5 - “You’re gonna need a bigger boat”
  21. 21. 22 #5 - “You’re gonna need a bigger boat” ü Jaws (1975)
  22. 22. 23 #5 –Appropriate Hardware for OpsManager •  Correctly sizing right at the beginning –  Consulting package –  Open a Commercial Support ticket •  Beware of changes in the initial estimates
  23. 23. 24 #6 - “Bond, James Bond”
  24. 24. 25 #6 - “Bond, James Bond” ü Dr. No (1962)
  25. 25. 26 #6 - Cloud/Ops ManagerAgents always initiate communications db.movies.find({“series”:”James Bond”}) .sort({“date”:1}) .limit(1)
  26. 26. 27 #6 - Cloud/Ops ManagerAgents always initiate communications Agent Hi, I am part of the 007 group. Server Go collect information about the hosts goldfinger, scaramanga and doctorno. Agent Here are their status and metrics for goldfinger and scaramanga. Agent doctorno is now permanently unreachable.
  27. 27. 28 #7 – “You've got to ask yourself one question: 'Do I feel lucky?’“
  28. 28. 29 #7 – “You've got to ask yourself one question: 'Do I feel lucky?’“ ü Dirty Harry (1971)
  29. 29. 30 #7 - Knowledge is King •  Use the right resources –  MongoDB.com/presentations –  Conferences: MongoDB Days/MongoDB World –  Consulting packages –  Commercial Support –  Free Support (Google groups and Stack Overflow) –  Training classes
  30. 30. Conclusion
  31. 31. 32 Take Away •  Remember 3 things from this presentation •  Make 1 change in your application or deployment •  Good return on your hour spent in this presentation
  32. 32. Questions… “You talking to me?”
  33. 33. Thank you and “May the Force be with you”

×