EPTS Survey results

1,091 views

Published on

Presentation of the Event Processing Survey prepared by the Use Cases Workgroup of the Event Processing Technical Society. Presented at the 6th EPTS Symposium at March 24, 2011

0 Comments
2 Likes
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total views
1,091
On SlideShare
0
From Embeds
0
Number of Embeds
2
Actions
Shares
0
Downloads
56
Comments
0
Likes
2
Embeds 0
No embeds

No notes for slide
  • (In)validate assumptionsTypes of users, actions, rules, queries, eventsCompare, constrast use cases
  • EPTS Survey results

    1. 1. Results of the Survey on Event Processing Use CasesMarch 24, 2011<br />Pedro Bizarro on behalf of the<br />Use Case Working Group<br />
    2. 2. 2<br />How are eventprocessingtechnologiesbeingused?<br />Classify scenarios,helpothersselect solutions<br />Inspire use ofevent processing<br />Food for thought forresearchers & engineers<br />
    3. 3. Earlier versions<br />2007: Kick-off – Problem: world of superficial use cases<br />2008: v1: 6 use cases, 54-questions (@4th epts)<br />2009: v2: 5 use cases  9 lessonslearned (@debs2009)<br />2011: v3: 30 use cases plentyofstatistics (today!)<br />3<br />
    4. 4. 4<br />24 questions<br />~13 minutes<br />30 use cases<br />EPTS, DEBS community<br />
    5. 5. Large Variety of Use Cases<br />5<br />Document workflow<br />Grid<br />Home energy<br />Patient discharge<br />Revenues and expenses<br />ETL in Telcos<br />Gas station networks<br />NYC transportation<br />Emergency management<br />Content authoring<br />Testing algorithms<br />…<br />
    6. 6. Industry background<br />6<br />Computer Software<br />Healthcare<br />Manufacturing<br />Retail and distribution<br />etc<br /><4% each<br />
    7. 7. Functional area<br />7<br />
    8. 8. Maturity level<br />8<br />52%<br />
    9. 9. Primary project drivers<br />9<br />
    10. 10. Data sources<br />10<br />Non“streaming”<br />“streaming”<br />
    11. 11. Destinations and actions<br />11<br />someautomation<br />but peoplestill stronglyin the loop<br />
    12. 12. Desired features<br />12<br />
    13. 13. Data models and data types<br />13<br />
    14. 14. Performance – input events per second<br />14<br />Not that much!<br />
    15. 15. Number of data sources<br />15<br />Data comes fromfew sources<br />
    16. 16. Number of “AI” models<br />16<br />3 in 4 don’t answeror don’t use<br />
    17. 17. Expected data size growth rate/year (in MB)<br />17<br />47% increase lessthan 10Gb/year<br />or not growing much<br />Hard to forecast<br />
    18. 18. Why did we use log scales?<br />18<br />Because we didn’t knowwhat to expect<br />
    19. 19. Enterprise capabilities<br />19<br />Cannot stop!<br />Not a big concern<br />
    20. 20. Implementation constraints<br />20<br />surprise<br />
    21. 21. The typical use case<br />In production to improve banking/utilities user services<br />Gets data from databases, files and message queues<br />Notifies people and other applications<br />Does correlations, joins and aggregations<br />Handles less than 1000 events/second (<10 sources)<br />Must run a specific CEP engine<br />Cannot stop<br />21<br />
    22. 22. The surprises (personal take)<br />Few telcos, transportation and logistics<br />Not from IT department<br />Not about lower TCO, faster deployments<br />“Non-streaming” dominate<br />Low throughput, few sources, little growth<br />Not about forecasting, predictions<br />Very few AI models<br />Not about security, encryption<br />22<br />
    23. 23. Membersandacknowledgements<br />Pedro Bizarro <bizarro@dei.uc.pt><br />ChristophEmmersberger <christoph.emmersberger@citt-online.com> <br />Thomas Ertlmaier <thomas.ertlmaier@citt-online.com> <br />Matthew Cooper <M.Cooper@eventzero.com> <br />Tina Groves <Tina.Groves@ca.ibm.com><br />Dieter Gawlick <dieter.gawlick@oracle.com><br />Brian Connell <brian@westglobal.com><br />23<br />
    24. 24. Q&A?<br />24<br />

    ×