Timing comparison for show commands

During this last time I’ve been doing some testing, making different kind of executions, in terms of delay_factor, max_loops, mixed together with send_config_set() and send_command().

My conclusion is three fold:

1. delay_factor should be tuned to improve interaction with the network;
2. if using config like commands, send_config_set() is better; if show-like commands, send_command() is better;
3. if send_config_set(), better off use it with cmd_verify=True.

I’ve performed some tests and it is true that when dealing with only config-like commands, send_config_set(cmd_verify=True) is an absolute winner.

But, when dealing with show-like commands, the picture is somehow different, at least for me.

Let me show you some results:

26 routers;
delayFactor=1;
maxLoops=5000;
device=nokia_sros;
commands=60 different show commands;
netmiko=3.4.0

And these are the code I’m using:

inText = [showCmd1, showCmd2, .... , shwCmd60]
if runType == 'config':
	output = conn2rtr.send_config_set(config_commands=inText, 
                                            enter_config_mode=False, 
                                            cmd_verify=cmdVerify, 
                                            delay_factor=delayFactor, 
                                            max_loops=maxLoops)
elif runType == 'show':
	output = ''
	for cmd in inText:
		output = output + '\n' + cmd + '\n' + conn2rtr.send_command(cmd, 
                                                                             delay_factor=delayFactor, 
                                                                             max_loops=maxLoops)

The results:

| send_config_set() | | send_command() – | – | --|– Data | verify=no | verify=yes | verify=n/a Count - Router | 26 | 26 | 26 showCommands | 60 | 60 | 60 Min - rxLines | 440 | 440 | 444 Average - rxLines | 5,211.77 | 5,212.73 | 5,216.85 Max - rxLines | 14282 | 14283 | 14287 Min - time(s) | 6.74 | 12.37 | 56.71 Average - time(s) | 13.31 | 17.81 | 60.81 Max - time(s) | 23.16 | 28.74 | 71.29

so, as you can see, the better performance in my case is not obtained with send_command() but with send_config_set().

It’s just a test that I wanted to share with you. I’ve repeated the tests several times, and the numbers are consistent.

Is it may be because I’m concatenating the results when using send_command() while send_config_set() gives me the result directly in the output variable? What is it that I’m doing wrong? Because show comands should better off treated with send_command(), isn’t it?

Regards!

Lucas

[ktbyers]

Okay, your numbers seem reasonably (i.e. make sense to me given how Netmiko works).

With the obvious caveat that I expect send_config_set() to generally not work and to not behave correctly for most show command operations (i.e. most people would run into issues using it).

[laimaretto]

@ktbyers , thanks for the reply. Just to make it clear, I fully understand that show commands should be used with send_command() while config commands with send_config_set(). This was just a test that called my attention because of the apparent delay when using shows + send_command() compared to shows + send_config_set().

So, bottom line, II’ll do as you suggest: 1) try the expect_string in the send_command() call; 2) Try the develop branch. I’ll let you know about the results. thanks!