>>5.Can be used for Load Testing?

 ok I'm gonna get up on a soapbox for a moment.

THIS is a bogus requirement.  If you want to say "simple" or
"rudimentary" loadtesting then I can accept it, but ANY tool maker
that claims their tool is good for both functional testing at the UI
level, and can ALSO function as a 'true' loadtesting tool either
doesn't understand loadtesting, or is just flat out lying to you.

The targets of these two things are vastly different, and the needs
are very different..  a functional test tool focusses on emulating a
single user driving the the way user's do, clicking on buttons,
inputting text in fields, etc..  it's looking at the function of the
UI and aims to do things like execute client side scripts, etc.   Such
a tool generally has to opeate at the UI level itself, not the
protocol level.  you need to emulate mouse clicks, typing, events like
mouseovers, and validate the client responds correctly.  You're also
looking to use different data for test for proper error handling, and
want things like the ability to do data driven tests etc.

A loadtesting tool aims at emulating the load upon a server or set of
servers ("system under test") from MANY users often over a period of
time, often emulating the peak load or a worst case load that may
involved thousands upon thousands of 'virtual users' hitting the SUT.
You don't care about client side code, and most often drive things at
the protocol level in order to allow a few physical systems to
generate those levels of load (something that is flat out impossible
if you are trying to drive multiple instances of a browser ).  You
need controls for scaling up the load, monitoring and controlling how
many virtual users are running at once, running multiple scenarios at
once (new users signing up, existing users doing transactions, etc)
AND the whole time you are doing that you want integrated montoring of
performance counters from all the systems under test (and the systems
generating the load).  You are looking to discover the capacity of the
system, or find problems in system design, scalability etc.

FYI: For this level of loadtesting I'd recommend either Microsoft
Visual Studio Team System (either Test Edition or Team Suite edition)
or a product like LoadRunner.  The former will run you several
thousand, the latter can easily run from tens to hundreds of thousands
depending on your needs.  (I've heard of a few opensource or free
tools that claim to do loadtesting, but I've yet to personally see one
that will match the capabilities of either of those.  I welcome input
from others,familiar with good low cost loadtesting tools, but lets
not derail the discussion too far shall we?)

Now, a tool like Watir (which I LOVE for functional testing of web
based apps), and many other tools that work at the browser level can
do very simple loadtests, e.g. concurrency tests, or simulate a
handful of users hitting a system at the same time, but that's about
it, simply due to lmitations of doing such things at the browser
level.  They also don't have the integrated systems to do the
performance monitoring of the SUT, which ends up being very critical
in terms of figuring out what part of the system failed)

If someone is trying to sell you on a tool they claim does both
things, it's time to nail then down on specifics..  such as how many
virtual users can a single client system emulate?   For a true
loadtesting tool the answer is typically along the lines of "it
depends on the complexity of the scripts (which truely it does) but
typicaly around 600-1000 vusers per system (or more)"  Ask them what
kind of instrumentation of the SUT and load rig is provided.. what
performance counters are monitored, if you are alerted when counters
pass critical thresholds indicating a performance bottleneck.  And if
they start trying to give you a song and dance about their tool being
more 'realistic' because it's actually driving a ui, turn around and
walk away because that kind of thing is 100% snake oil! .. (because
the servers are responding to HTTP requests, and frankly there is NO
WAY for the servers to tell if the requests came from a tool like
VSTS, or a browser).

sorry for the tirade, but his is a pet peeve of mine.  You shouldn't
try to do loadtesting with a Functional test tool, any more than you
should use a wrench as a hammer.. (it might work sort of, in a pinch,
but you are not very likely to get the results you need)


On Jan 2, 12:43 am, Parul <parul...@gmail.com> wrote:
> Hi all,
> Can anyone help me in giving answers that which of these are supported
> by watir and which are not.Below are some project requirement:
> 1.Should support all type of Fonts.
> 2.Sturts2
> 3.Overlay Pages
> 4.Help Bubbles
> 5.Can be used for Load Testing?
> 6.Command Line, Batch Mode, Versioning
> 7.Tidal Invocation
> 8.Data Wizard(data bank)
> 9.Navigation should support by two ways
>     1)By programmatically and clicking buttons [ Index wise, position
> wise]
> 10.Fault Tolerance
> 11.Behaviour on different browsers
> 12.Exclusion Rules (test cases within suite)
> 13.Reporting
> 14.Interoperability of tools
> scheduling mode and batch Mode
> Centralization(Clinet and Server)
> It should support share module (Library of function developed in VBA)
> E-mail notification for results
> Synch of with VSS
> Script Portability
> Run against different version, env, releases
> Thanks
> Parul
You received this message because you are subscribed to the Google Groups 
"Watir General" group.
To post to this group, send email to watir-general@googlegroups.com
Before posting, please read the following guidelines: 
To unsubscribe from this group, send email to 
For more options, visit this group at 

Reply via email to