Computing Reviews
Today's Issue Hot Topics Search Browse Recommended My Account Log In
Review Help
Search
Serverless computing: economic and architectural impact
Adzic G., Chatley R.  ESEC/FSE 2017 (Proceedings of the 2017 11th Joint Meeting on Foundations of Software Engineering, Paderborn, Germany, Sep 4-8, 2017)884-889.2017.Type:Proceedings
Date Reviewed: Jun 22 2018

The concept of virtual machines has liberated--elegantly and very successfully-- information technology (IT) from hardware constraints. The next step in virtualization now concerns the very programs themselves: in “serverless” computing, cloud providers like Amazon, Microsoft, Google, and IBM allow developers to deploy single functions to a cloud infrastructure rather than whole servers. The cloud provider then only bills for the central processing unit (CPU) time consumed when a particular function is really executing. The name stems from the fact that programmers no longer control any “server process” that is taking care of listening to events (which “wake up” the function), starting up or dispatching functions, or reserving sufficient processing capacity.

The first half of the paper provides a lucid and easy-to-follow introduction to this concept, together with a short but fairly trivial cost comparison of a classical deployment model (where one has to pay for the running hours of a virtual machine irrespective of whether something is running or not) to a serverless model (billed by the actual execution time of invoked functions).

It is the second part of the paper that makes it worthwhile. Based on two nontrivial real-life case studies (one of which employs almost 200 serverless functions), the authors highlight the differences in application architecture brought about by a switch to serverless computing. Using sequence diagrams, the authors carefully show how an (admittedly more complex) execution flow of an application composed of serverless functions may result in significantly (circa 60 percent) reduced cloud operation costs.

The paper ends with a good and concise discussion of some selected opportunities (for example, removing barriers to bundling or versioning) and shortcomings (for example, invocation latency, short execution time limits in the area of minutes) of serverless computing.

This is a timely and highly recommended read, not only for solution and software architects but also for application and information technology (IT) managers who want an introduction to this domain.

Reviewer:  Christoph F. Strnadl Review #: CR146107 (1809-0506)
Bookmark and Share
  Reviewer Selected
Featured Reviewer
 
 
Distributed Programming (D.1.3 ... )
 
 
Patterns (D.2.11 ... )
 
 
Programming Environments (D.2.6 )
 
Would you recommend this review?
yes
no
Other reviews under "Distributed Programming": Date
Topics in distributed algorithms
Tel G., Cambridge University Press, New York, NY, 1991. Type: Book (9780521403764)
Sep 1 1992
Interacting processes
Francez N., Forman I., ACM Press/Addison-Wesley Publ. Co., New York, NY, 1996. Type: Book (9780201565287)
Jan 1 1997
Verification of sequential and concurrent programs (2nd ed.)
Apt K. (ed), Olderog E., Springer-Verlag New York, Inc., Secaucus, NJ, 1997. Type: Book (9780387948966)
Feb 1 1998
more...

E-Mail This Printer-Friendly
Send Your Comments
Contact Us
Reproduction in whole or in part without permission is prohibited.   Copyright 1999-2024 ThinkLoud®
Terms of Use
| Privacy Policy