Article: A Formal Performance Tuning Methodology: Wait-Based Tuning

Posted by Abel Avram on Oct 06, 2008 02:33 PM

Community

Architecture,

Java

Topics

Performance & Scalability

Tags

Performance Tuning ,

Performance Evaluation

In this article, Steven Haines talks about web application performance tuning which used to be more of an art than science. He proposes a method called wait-based tuning, making the entire process more measurable and, consequently, more scientific.

Read: A Formal Performance Tuning Methodology: Wait-Based Tuning

Steven explains why web application performance tuning is difficult to perform. A client request may go through several tiers suffering delays at different levels before a response is returned:

Input can come in the form of an HTML browser, a thick client, a mobile device, or a web service, which can pass through servlets running in one of a dozen different architectures or a portal container, that in turn may call enterprise beans, external web services, or delegate processing to a business rules engine. Each of these components may then interact with a content management system, a caching layer, a plethora of databases, and legacy systems.

The complexity of tuning today's web applications can be addressed by analyzing the architecture of the application and defining a series of wait-points marking the path traversed by a client request until it becomes a response that reaches back to the client. There are 2 types of wait-points, according to Steven:

1. tier-based, marking the transition between tiers
2. technology-based, mostly related to the usage of underneath infrastructure

Steven says that having "a set of wait-points identified, the tuning process is implemented by opening all tier-based wait-points and external dependency pools, generating balanced and representative load against the application, and tuning backwards, or tightening wait-points to maximize the performance of a request’s weakest link, but without saturating it."

1. Back to top

   A fresher look at queuing theory ..

   Oct 7, 2008 5:42 AM by muthu kumaran

   Nice article applying less-known potentials of queueing theory and queuing models. QoS-intensive telecom network software use this extensively in capacity prediction across various version releases. Resource wait-time is one of the key tuning ingredient in most of telecom services software/middleware and calls for very careful engineering to optimize the performance.
   
   Good that the perspective is neatly extrapolated to web application model through this article.

   Reply

2. Back to top

   Re: A fresher look at queuing theory ..

   Oct 9, 2008 2:25 PM by d taye

   how different is this from what's described here:
   

   
    www.infoq.com/articles/SEDA-Mule
   

   Reply

3. Back to top

   Re: A fresher look at queuing theory ..

   Oct 9, 2008 3:55 PM by muthu kumaran

   SEDA is a formal architectural paradigm which emphasizes what is stated in the above article. Instrumenting the code to carry wait-time analysis (monitoring request wait-times in various queues through critical application flow, resource monitoring) and implementing a graceful degradation are all design paradigms are facilitated by SEDA through customization mechanisms. This article seems to discuss about the the approach which is apparently facilitated (with some customizations) by frameworks such as Mule

   Reply

4. Back to top

   Good one

   Nov 18, 2008 10:55 AM by vijay sasi

   Good article ! Just one thing to clarify I couldn't relate the caching any where in the application tier.

   Reply

• Older >