Presentation is loading. Please wait.

Presentation is loading. Please wait.

Behavior Isolation in Enterprise Systems Mohamed Mansour

Similar presentations


Presentation on theme: "Behavior Isolation in Enterprise Systems Mohamed Mansour"— Presentation transcript:

1 Behavior Isolation in Enterprise Systems Mohamed Mansour mansour@cc.gatech.edu

2 Feb14, 20072 Client 1 Message queue Travel Industry Example Client 2 Client 3 clearinghouse Airlines GDS

3 Feb14, 20073 GDS Scale Mission critical environment  24/7 11.5 million queries/days 2-16 seconds processing time ~10GB data set, 20% annual increase 8 updates per day, moving to seamless updates Message queue GDS

4 Feb14, 20074 Effect of Request Stream

5 Feb14, 20075 Why We Care? Business  Consumer Loyalty  Violates contractual agreements Technical  Occurs even in highly engineered systems  Can cause ripple effects

6 Feb14, 20076 Lets Just Fix it! Difficult to identify root cause  Constant data changes  Request stream dependency Sometimes can’t fix root cause  3 rd part libraries  Interactions with OS, and H/W caches  Complex code base

7 Feb14, 20077 I(solation) Queue Dynamic management of message streams Correlate message sequences with server behavior  Learning phase Isolate undesired sequences  Control phase Evaluation metrics  Quality of Information metrics (QoI)

8 Feb14, 20078 Learning Phase Use online learning methods  Statistical correlation [ICSOC 06]  HMM [GIT-CERCS-06-11] Behavior Model  Associate undesired behaviors with certain input patterns

9 Feb14, 20079 Control Phase Observe input message sequence Control sequence dispatched to each server to maintain QoI  Dispatcher  Reordering messages in queue

10 Feb14, 200710 I-Queue Applied to Worldspan Pricing Engine  Affects customer relations  Possible impact on consumer experience – less options Objective: return maximum number of alternate fares Problem  Variable number of alternate fares for same query  Root cause unknown

11 Feb14, 200711 Establishing Behavior Model Heuristics point to query geographies  Geography based on From/To city pair, e.g. East Coast to EU  Fare data stored in disk files separated by geography Use geo-locality as our predictor  Goal: improve geo-locality

12 Feb14, 200712 Modified Queue Dispatcher Dispatcher maintains server execution history Request routed to an available server with matching geography Message queue GDS

13 Feb14, 200713 Evaluation Used real traces from Worldspan  Set of about 1800 requests  20% process in 16 seconds Geography extracted from messages  Hand-coded mapping from city pairs to geography code Processing times measured using Worldspan servers  Completely static environment Simulations to measure geo-matching  Compare different isolation points

14 Feb14, 200714 Improvement in Geo-locality Matching improves 6 times for min. farm size Matching can improve further by adding more servers

15 Feb14, 200715 Choosing the Right Metrics to Monitor Min. of 28 servers to avoid queuing delays Geo-match increases with more servers Queuing delay is not the best metric to monitor

16 Feb14, 200716 Future Directions


Download ppt "Behavior Isolation in Enterprise Systems Mohamed Mansour"

Similar presentations


Ads by Google