This post is password protected. To view it please enter your password below:

Password:

“c:\Program Files\Log Parser 2.2\LogParser.exe” “select time-taken into aaa.gif from ex091116.log where [cs-uri-stem] like ‘%myurl%’ and to_time([time]) between to_timestamp(’12:15:00′, ‘hh:mm:ss’) and to_timestamp(’20:20:00′, ‘hh:mm:ss’)” -view

It seems that port 139 is the port used on the remote machine.  There were no instances of traffic to the remote machine that did not not use port 139, so that seems to be a good sign.

This also agrees with most of the blogs I found that mention only port 139.

I’ll update this blog if I find otherwise.

You  can check the port by using telnet <remote machine> 139.  It should go to a black screen, and not show any connection error.  There are still permissions to deal with on the remote machine though.

Here’s how it can be done:
1. telnet to the web server, usually it is listening on port 80.

telnet 10.1.1.1 80

2. Send the HTTP request to get a page.  Remember that there is a blank line after the GET line.  The /mydir/Default.aspx is the path to the file on the web server. In most cases this will be just /index.html or something like this.

GET /mydir/Default.aspx HTTP/1.0
3. You should get an HTML page back and then the session disconnects.

http://www.storageperformance.org/home

Most sites say to be wary of benchmarks because each SAN is different, and what it is used for needs to be taken into account.

While this is true, it is still good to have some numbers to compare to when building a new SAN .

http://www.microsoft.com/technet/prodtechnol/WindowsServer2003/Library/IIS/2228ff79-0838-4143-8680-e38c452dc243.mspx?mfr=true

Basically it says to “Run Iisapp.vbs with no parameters”

What I need to know before I can start:
1. What is the required transaction for an hour (TPH)
2. What is the longest time that a transaction should take (Duration)

The calculation:
a) Calculate the transactions per second (TPS) – TPH / 3600
b) Calculate the Pacing time, this is a bit longer than the Duration, to allow for unexpected delays in processing.

If the transaction is a quick one, (less that 5 seconds) I typically add on 2 seconds for the pacing.  If it is a longer transaction (more than 5 seconds) I’ll add on 5 to 10 seconds.

c) Calculate the level of concurrency, that is how  many transactions will be running at the same time
TPS x Pacing = Concurrency

Concurrency is the number of virtual users that you’ll need.

Lets look at an example

Login, takes no more than 2 seconds to run and we require 7200 for the peak hour (nice numbers to make the math easier).

a) TPS = 2    (7200 / 3600)
b) Pacing = 4 seconds (2 seconds Duration + 2 seconds extra)
c) Concurrency = 8 (2 x 4)

So 8 virtual users are required to produce this load.

Now lets work backwards to check the calculation

If I have 1 user performing a transaction every 4 seconds I will get (3600 / 4) 900 transactions in an hour.

If I have 8 users doing the same I will have (900 x 7200 transactions in one hour.

This is the number that I want.

Some reality checks:

During the run, check the TPS for the transaction matches what you calculated.  Multiply the TPS you see in your load tool by 3600 to check it gives you your required hourly count.  If it is a bit under, make sure you remember any ramp up, as it will impact you TPS average.

After your first run, make sure you have the right total number of transactions.

What about ramp up?

I generally treat ramp up separately from the one hour of load.  So I’ll add 15 minutes of ramp up to the start of my test (I call this the Ramp Up Period), meaning the whole test runs for 1 hour and 15 minutes.  I typically don’t worry about ramp down.
Ramp up is simple to calculate

Ramp Up Interval = Ramp Up Period / Virtual Users

From our example:
Ramp Up Interval = 900 seconds (15 minutes) / 8 (Virtual Users)

Based on this a new user should start every 112 seconds

This means the users will be fully ramped up before the 15 minutes, due to rounding, but it doesn’t matter that much in the big scheme of things.

Once again, to check the math, divide the Ramp Up Period by the Ramp Up Interval
900 / 112 = 8.036 which is close enough to 8.

Be sure to check as the script runs that all users are up and running after 15 minutes (or whatever your Ramp Up Period is).

Sometimes you need to check if a port is open for receiving on another computer.

It could be that you want to check if a program has opened the right port, ready to receive, or as it was in my case, to check if a firewall had been opened up (burned) correctly to allow a connection through on a specific port.

A good way to do a simple check is with telnet.

From a command prompt run

telnet <machine name or IP address> <port>

For example

telnet 10.10.10.32 2837

If the port is open, the screen in DOS look like:

If the port is not open you see something like:

http://motevich.blogspot.com/2008/08/bugs-in-loadrunner.html