preferably non-parametric signs that can be indentified
directly from profiles. Another challenge in working
with real web APIs is probing overhead. Web APIs en-
force strict rules about the frequency and types of API
access, e.g., 2 accesses per second per user [7].
Our current research closed the loop between cloud
applications and web APIs, By providing more mean-
ingful profiles of APIs, cloud applications will be able to
control their internal system in a more effective way. It is
worthwhile to explore how to build a robust cloud appli-
cations by using profiles of APIs recovered by BSS. For
example, a cloud application may dynamically dispatch
requests to different APIs based on their profiles to avoid
busy hours of an API.
In conclusion, we proposed research on profiling third
party web APIs using BSS techniques. Using data col-
lected outside of an API provider’s system, we are able
to “look in” at detailed workload profiles. In early re-
sults, we used ICA to recover accurate profiles. We also
showed that our workload profiles were helpful, provid-
ing insight into design of tested services.
References
[1] 1998 World Cup Workload. http://ita.ee.lbl.gov/
html/contrib/WorldCup.html.
[2] Apache zookeeper. http://zookeeper.apache.org/.
[3] Carbon-aware energy capacity planning for datacenters.
[4] Redis. http://redis.io/.
[5] Spark: Cluster computing with working sets.
[6] Stackoverflow question 10986702:is zookeeper appropriate for
object caching? http://stackoverflow.com, 2012.
[7] Programmable web: Mashups, apis and the web as a platform.
http://www.programmableweb.com, 2013.
[8] Stackoverflow question 1479442:real world use of zookeeper.
http://stackoverflow.com, 2013.
[9] M. Attariyan and J. Flinn. Automating configuration trou-
bleshooting with dynamic information flow analysis. In USENIX
OSDI, 2010.
[10] P. Barham, A. Donnelly, R. Isaacs, and R. Mortier. Using magpie
for request extraction and workload modelling. In OSDI, 2004.
[11] M. L. C. Stewart and K. Shen. Empirical examination of a col-
laborative web application. In IEEE International Symposium on
Workload Charactization, Sept. 2008.
[12] P. Comon and C. Jutten. Handbook of Blind Source Separation:
Independent Component Analysis and Applications. Academic
Press, 1st edition, 2010.
[13] T. Everts. An 11-step program to bulletproof your site against
third-party failure. http://blog.radware.com/, 2013.
[14] T. Everts. Case study: Understanding the impact of slow
load times on shopping cart abandonment. http://blog.
radware.com/, 2013.
[15] R. Fonseca, G. Porter, R. H. Katz, S. Shenker, and I. Stoica. X-
trace: A pervasive network tracing framework. In Proceedings
of the 4th USENIX conference on Networked systems design &
implementation, 2007.
[16] A. Gandhi, Y. Chen, D. Gmach, M. Arlitt, and M. Marwah. Min-
imizing data center sla violations and power consumption via hy-
brid resource provisioning. In IGCC, 2011.
[17] A. Gandhi, S. Doroudi, M. Harchol-Balter, and A. Scheller-Wolf.
Exact analysis of the m/m/k/setup class of markov chains via re-
cursive renewal reward. In ACM SIGMETRICS, 2013.
[18] A. Gandhi, M. Harchol-Balter, R. Raghunathan, and M. A.
Kozuch. Autoscale: Dynamic, robust capacity management for
multi-tier data centers. ACM Transactions on Computer Systems
(TOCS), 30(4):14, 2012.
[19] I. n. Goiri, W. Katsak, K. Le, T. D. Nguyen, and R. Bianchini.
Parasol and greenswitch: Managing datacenters powered by re-
newable energy. Mar. 2013.
[20] X. Gu and H. Wang. Online anomaly prediction for robust clus-
ter systems. In Data Engineering, 2009. ICDE’09. IEEE 25th
International Conference on, pages 1000–1011. IEEE, 2009.
[21] Y. Kwon, S. Lee, H. Yi, D. Kwon, S. Yang, B.-G. Chun,
L. Huang, P. Maniatis, M. Naik, and Y. Paek. Mantis: automatic
performance prediction for smartphone applications. In USENIX
Annual Technical Conf., 2013.
[22] G. Lu, J. Zhan, H. Wang, L. Yuan, and C. Weng. Powertracer:
Tracing requests in multi-tier services to diagnose energy inef-
ficiency. In Proceedings of the 9th international conference on
Autonomic computing, pages 97–102. ACM, 2012.
[23] D. J. C. MacKay. Information Theory, Inference & Learning Al-
gorithms. Cambridge University Press, New York, NY, USA,
2002.
[24] H. Nguyen, Z. Shen, X. Gu, S. Subbiah, and J. Wilkes. Agile:
elastic distributed resource scaling for infrastructure-as-a-service.
In Proc. of the USENIX International Conference on Automated
Computing (ICAC13). San Jose, CA, 2013.
[25] A. Peters. Why loading third party scripts async is not good
enough. http://www.aaronpeters.nl/, 2011.
[26] K. Shen, A. Shriraman, S. Dwarkadas, X. Zhang, and Z. Chen.
Power containers: An os facility for fine-grained power and en-
ergy management on multicore servers. In ACM ASPLOS, 2012.
[27] K. Shen, M. Zhong, S. Dwarkadas, C. Li, C. Stewart, and
X. Zhang. Hardware counter driven on-the-fly request signatures.
In ACM ASPLOS, Mar. 2008.
[28] C. Stewart, A. Chakrabarti, and R. Griffith. Zoolander: Effi-
ciently meeting very strict, low-latency slos. In Int’l Conference
on Autonomic Computing, 2013.
[29] C. Stewart, T. Kelly, and A. Zhang. Exploiting nonstationarity for
performance prediction. In ACM European Systems Conference,
Mar. 2007.
[30] C. Stewart, K. Shen, A. Iyengar, and J. Yin. Entomomodel: Un-
derstanding and avoiding performance anomaly manifestations.
In IEEE MASCOTS, 2010.
[31] E. Thereska and G. R. Ganger. Ironmodel: Robust performance
models in the wild. ACM SIGMETRICS Performance Evaluation
Review, 36(1):253–264, 2008.
[32] TripAdvisor Inc. Tripadvisor reports fourth quarter and full year
2013 financial results, Feb. 2014.
[33] H. Tsukayama. Facebook outage takes down gawker, mashable,
cnn and post with it. http://21stcenturywire.com/,
2013.
[34] Z. Zhang, L. Cherkasova, A. Verma, and B. T. Loo. Automated
profiling and resource management of pig programs for meeting
service level objectives. In Proceedings of the 9th international
conference on Autonomic computing, pages 53–62. ACM, 2012.
6