VPCS and VPCS multi-host

I have seen a few requests for help about VPCS on the GNS3 forum lately, and some confusion about the VPCS multi-host option and how it is different to the regular GNS3 VPCS host. So if you are confused, or just want to know more about how you can use VPCS in your GNS3 topologies, read on.

Single instance VPCS

There are two ways you can run VPCS – by far the easiest is to add a VPCS host to your topology. By dragging one in from the End devices pane. Once you have added the host, you can start the VPCS process by selecting the VPCS device, then choosing Device | Start (the Device menu can be also accessed by right-clicking on a device).

Once started, a console session can be established with the Virtual PC by choosing Device | Console.

And from here, you can explore VPCS options using the ?


To get stated, you’ll need to set an IP address, like this


…which will set the IP address for the VPC to with a mask of 255.255.255,0 and a default gateway of

And the next most important command you’ll want to learn is:


…which will of course you MUST do before saving your project within GNS3 – GNS3 will NOT save your VPCS (or your Cisco router configs) for you if you forget.

From this point, you’ll want to explore the ping and trace commands. But there are other commands worth checking out like:

show ip 


set dump detail

…to see a summary of packets sent and received by this VPC. (Use set dump off to turn it off)

You can then add another Virtual PC and another – as many as you like.

But there is another way of using the Virtual PC Simulator that gives you even more options – you may have seen it on your Tools menu. VPCS multi-host

VPCS multi-host

Now VPCS multi-host is considerably harder to set up than the normal VPCS. In fact, you have to set it up the way VPCS used to have to be set up in GNS3 v0.x – using either the cloud or host device (both are essentially the same)

To set up a VPCS multi-host, drag either a cloud or host device onto your topology.

Next, select the device, and choose Device | Configure. This brings up the Node Configurator (hopefully soon this will change name to the Properties dialogue).

From here, choose the UDP tab.

Initially, the settings should be just perfect for VPCS[1] – with the Local port setting at 30000, the Remote host set to, and the Remote port set to 20000, so click the Add button, then click OK.


You can now use the connector tool to connect to this nio_udp:30000: interface you have created to another device (make sure you select the nio_upd interface)


Now there isn’t much point using the VPCS multi-host option if you don’t want more than one host, but when you add your second host, you’ll have to be a little more careful at the step where you add the UDP settings under the UDP tab. For the second VPCS host, you need to use ports 20001 and 30001. And for the third VPCS host, ports 20002 and 30002 and so on. You can configure up to nine VPCS hosts in this way.

TIP:RedPoint2 When configuring the UPD ports for the second and subsequent hosts, you might find it easier to click the Add button multiple times until the port number you need for that host number appears in the box. You can then delete the others, or leave them there until you have built your topology – GNS3 will prune any extra unused ports for you when you quit the project!

Once you have got your VPCS hosts connected to your topology, you can choose the Tools | VPCS multi-host option to spawn a session with your hosts. And here is where the first big difference between the regular VPCS hosts and the VPCS multi-host option becomes apparent.

No matter how many VPCS multi-host you configured, only one window opens – and the prompt is slightly different; it will show:


Now the commands for VPCS multi-host are exactly the same as they are for the regular VPCS, except that all of your Virtual PCs are in one window – to access Virtual PC #2, you will have to enter the digit 2 on a command line by itself and hit <Enter> The prompt will change to


You can navigate to any of the 9 Virtual PCs in the same way – press digit <Enter> – as explained in the now slightly expanded help screen.

What’s the point? Why go to all that trouble to set up VPCS multi-host?

There are some simple advantages:

  • You only have to remember to save one file – a single save command saves the config of all your Virtual PCs (up to nine)
  • When you want to ping the same IP address from different hosts, simply change focus to a different Virtual CP, and use the up-arrow key to retrieve commands in your command history.
  • When you use the set dump command, you can see the results, even if the dump is for a different Virtual PC than the one in focus.
  • A single window is easier to mange than multiple windows, especially if you are using a totally Windowed terminal application that doesn’t support tabs (Putty for example).
  • Only a single instance of VPCS is spawned for all (nine) Virtual PCs, which means the amount of RAM and CPU required by VPCS is reduced by a factor of the number of single-instance VPCS you WOULD have used :)

But the most significant difference between Single instance VPCS and VPCS multi-host is possibly one you may never use, but incredibly powerful just the same.

With VPCS multi-host you can load a script file that tests your configuration from multiple points. There is no easier way to explain this than to show you a script checking one of my GNS3 WorkBench exercises. You’ll have to watch the video to see, but in it you’ll see multiple tests being carried out from multiple Virtual PCs, and the set dump feature used on a “remote” Virtual PC (the http://www.example.com host) to observe that IP addresses have been NATted to the correct IP address.

The exercise used in the demonstration above was my ICND1 Readiness Test, part of the GNS3 WorkBench.

So the choice is yours – use either the regular Single instance VPCS, or configure your hosts to run in a single VPCS multi-host instance – you might even get adventurous and start writing your own scripts.

For more information on how to run the VPCS multi-host, check out my VPCS Tutorial


Posted in GNS3, GNS3 WorkBench, vpcs | Tagged , , , | 3 Comments

Cisco have a new ACI lab on dcould – here’s how to access it

If you sell/buy/use Cisco products, and you want to practice configuring them, then Cisco’s dcloud is a great resource.  And they have just added another ACI lab: Cisco Application Centric Infrastructure with shared Physical Fabric.

If you already know about dcloud, then go ahead, check out the new lab.  But if you are new to dcloud, then this tutorial will get you going.  You will need a cisco.com login before proceeding. If you don’t have one, you can register here.

Start ny accessing dcloud at https://dcloud.cisco.com


The first time you access the site for a given browser, you will have to choose your location, and of course log onto the site using your cisco.com logon details (available to anyone who signs up).

TIP:RedPoint2  Don’t use Internet Explorer for this lab if you want to access the remote desktop PC in full screen mode.


…and of course you’ll have to accept the Cisco dcloud User Agreement and fill out some Profile details if this is the first time for a particular theatre.


Once you hit the Home page, you might like to explore some of the training videos, but for me I want to go straight to the new ACI lab, so I’m going to open the Application Centric Infrastructure section and click More Information, because that is where I’ll download the pdf guide to the lab, and also find a Schedule option there anyway.


By default, your schedule will start at the next quarter hour time slot, but you can click Now is you are impatient. I was! And you can book the lab continuously for up to five days,


You’ll need to fill in some details about why you want the lab, and you session will begin.  It will take up to 15 minutes for the lab to prepare, because what is now happening is that the ACI simulator and a whole swag of Virtual Machines are being loaded up for your use. [Edit: this particular lab doesn’t use the simulator – it gives you access to real ACI kit with your own login]

Now while this is happening, it’s worth thing a little about what Cisco is doing in the background to make this happen.  If you are reading this, then you probably have some knowledge of SDN and Orchestration – and you are now watching it in action.

Once your session is ready, your dashboard will change to:


Now the first thing you will need to do is View your demo, and when you do, you will see a topology with a glowing workstation inviting you to click it! But I want to draw your attention to a couple of other sections on this screen before you launch into your lab:


Check out the DocumentationSession Details, and the Servers sections under your Topology section. Get used to working your way around these because later I’ll refer you to the Session Details section later.

Note:RedPoint Note the time remaining – if you actually manage to schedule more than five days, don’t believe it. Your session WILL end exactly five days later, even if you see that there are three hours to go.

But now it’s time to launch the lab.  The lab is conducted via a “Jump Box” – a session to a remote PC, and clicking on the Wkstn1 icon and selecting the Remote Desktop option. Note there are options to Power Off and Reboot the remote PC if you should get trouble.

However, the first thing you will probably want to do after launching the remote desktop is to expand the window that opens into full screen mode – some of those ACI configuration screens need a LOT of real-estate.

And this will lead you to your first problem: The screen doesn’t actually go full size – but not to worry, there is a solution (unless you still went ahead and used Internet Explorer in spite of my earlier warning).

After you have maximised your desktop, click in the grey area to the side of the login screen and choose Reload Page.


Your session will now be in full screen.  Log in using the user credentials given (dcloud\demouser password C1sco12345) and begin your lab.  You did download the pdf of the lab earlier didn’t you? If not, you can go back to your home page and click on the documentation section.  You’ll need it to find out the login details.

I’ll leave you to enjoy the lab yourself, but before I finish, I want to suggest that running a remote desktop session in a web browser may not give you the best user experience you can get from this site.

A better way to get to the remote desktop is to use Cisco AnyConnect. You can do this one of two ways.

The better way is to download and install the Cisco AnyConnect Client, but a good alternative is to use a SSL connection via your web browser.  In either case, you’ll have to return to your dcloud Home  page and select the Remote Session  section to get the login details you’ll need to make the connection.  This is also the place where you’ll find the link to the SSL login as well.


Note that you can share your login with up to 15 other users, each user has the same password, but a slightly different username of the form vNNNuser1..vNNNuser16

If you choose to use the web based SSL connection, you’ll have to have the latest versions of Java and the wind will need to be in the right direction and the gods smiling upon you. For me, the easier option is to install the VPN Client.

Once your connection is established, all you need to do now is open your Windows Remote Desktop client, and connect to the ip address shown next to yoru remote desktop in your Topology section.


Once you have saved your credentials and favourite screen settings in your Remote Desktop Client, accessing dcloud is a breeze forevermore!


But it doesn’t necessarily stop there.  I find I get the best performance via the RDP connection using my own RDP client and my own Cisco AnyConnect client, but there is yet another way.

Look at the IP address that you see when you access the AIPC in the remote lab:


So long as you are connected via your AnyConnect connection, and you have the password, you can log into this address on your own local browser. I actually find this less satisfactory than the RDP session, but it might suit you.  And I find Chrome more reliable than Safari.

remoteBrowser[Update: 2015-06-30]

Once you have logged into your lab, you might find it tedious swapping to and from your remote session to read the lab guide that you downloaded earlier – especially if you are operating in a full-screen remote desktop session.

So if this bothers you, open a browser in your remote session, and browse to https://dcould.cisco.com and log in again in the remote session – and then you can download the .pdf of your lab instructions to your remote PC and keep it all in one place.

[End of update: 2015-06-30]

Enjoy your lab!


Note:  This is the first of a few articles I’m writing that relate to ACI. Watch this space for a a multi-stage tutorial on ACI

Posted in Cisco, Cloud, Cloud computing, Data Center, Data Centre | Tagged , , , , , , | 4 Comments

Best after sales support I’ve ever had. Thumbs up for Everki

So here is my backpack. I’ve cut the straps and destroyed it. It was the best backpack I’ve had in 20 years of lugging around a computer



So why did I cut the straps?

To answer that, have a look at another photo.


In this one you can see that the zipper on my backpack has broken :(

So what has that got to do with cutting the straps?

And to answer that, I’ll have to tell you about the support I got from Everki when I told them about the broken zipper.

I have a problem with my Beacon laptop bag. There is part of the zip that seems to have got out of step and missed a bit – and one of the fasteners has become detached. See attached photos.

Is this covered by my “Lifetime Warranty”? If so, what do I do next? I *THINK* I bought it directly from your online store.

BTW – I LOVE the bag – it has been the best bag I have EVER had (and I’ve been carrying a laptop almost daily since 1995)

What followed was a series of exchanges asking me to return the bag, me explaining that I couldn’t because I was travelling so much and so on – every step of the way I got fantastic understanding from Andrea and her supervisor to the point where I didn’t have to return the bag, just a) cut off the serial number, and b) cut the straps.

Here is the result:

2015-05-17 16.07.04



A brand new bag shipped to me in Australlia at no expense to me.

Thank you Everki, I love your product and I love your service even more. And a special thanks to Andrea who patiently took me through the process even when I send the wrong pictures!







Posted in GNS3 WorkBench

Pi day

Just a heads up for tomorrow. I’m sure you all realise that tomorrow is Pi day – because if you write the date tomorrow in the American format (3/14) you get the first three digits of Pi.

But do you realise that Pi day this year has extra special significance, because at 3/14/15 at 9:26:53.59, the date/time combination will yield the first 12 digits of Pi (3.14159265359)?

This is a unique event in the history of recorded time so will never happen again – set your alarms, but you’ll only have 1/100 sec to enjoy the full 12 digits. Of course you can still bask in the knowledge that the whole day is also uniquely giving the first 5 places!

Posted in GNS3 WorkBench

Open What?

I was ask by some students to describe the difference between the “Open” standards associated with SDn (Software Defined Networking)

So I did this little presentation for them


Posted in GNS3 WorkBench | Tagged | 2 Comments

GNS3 and IOU remote server install on Centos 6.5

GNS3 and IOU remote server install on Centos 6.5.

Posted in GNS3 WorkBench

Subnetting without tears

Forget the bitwise operations. Here’s how I do subnets.  In my head!  Any subnet, any mask in 30 seconds or less.

1.    First – credit where it is due.  My method has been shamelessly stolen from Wendell Odom’s CCENT/CCNA ICND1 Official Exam Certification Guide. If you want to learn more, like how to use magic numbers to calculate the number of subnets for a given mask, read his books.
2.    The whole secret around subnets lies in:
a) the subnet mask
b) knowing your 2x, 4x, 8x, 16x, 32x, 64x and 128x times tables as far as 256
Most people can handle 2x, 4x and (at a pinch) 8x.  32x, 64x and 128x only have 8, 4 or 2 items before you hit 256, so that leaves you your 16x tables to learn.

Here’s the deal:

Look at the “interesting” number in the subnet mask.  That is, the number that is NOT 0 or 255.  In your case:
Your “interesting” octet is octet 4 ie 240.
Subtract this number from 256 – you’ll get 16 (256-240=16) in this case.

The key concept here is that your subnets will be in groups of size 16 

Back to

256-240=16.  Odom calls this the “magic number”.  I’ll stick with his definition.

The key to understanding subnet is this “magic number”.  Read the following carefully:

If the “magic number” is 16, then

  • ALL of your subnet numbers will be multiples of 16, starting at 0
  • The broadcast address on any subnet is simply the next subnet address minus 1 in the last octet
    • ie All of your broadcast addresses will be 1 less than a multiple of 16. (15, 31, 47…)
  • The first IP address on any subnet is simply the subnet address plus 1 in the last octet (1,17,33…)
  • The last IP address on any subnet is simply the broadcast address minus 1 in the last octet. (14,30,46…)

Now read that last section again until it sticks.

In your case, where the magic number is 16, the subnet MUST be one of the following:

There you go – that’s your 16x tables up to 256. Told you to learn this didn’t I?  By the way, your 32x tables is just every 2nd number from the above.,  64x tables every 4th number and 128x tables every 8th number.

Note that the last entry before 256 is the same as the subnet mask’s “interesting octet”.  This is ALWAYS the case.

Back to again.

Now your “interesting octet” is the 4th octet, which for your IP address is 15.

  • Your subnet number MUST be less than 15, but one of the multiples of 16 listed above.  Clearly 0 is the only multiple of 16 less than 15.  So your subnet is
  • The next subnet would be ie the NEXT multiple of 16.
  • The first address on your subnet will be =
  • The broadcast address will be the NEXT subnet minus 1 ie =
  • The last address will be one less than the broadcast address: =

So the address you started with is the broacast address for its subnet =

Here’s another example:
Magic number=256-248=8
The “interesting octet” is the last octet, so I focus on 219
Recalling my 8x tables around 219

  • I can see that my subnet number must be .216 – the closest multiple of 8 to 219 that is not greater than 219.
  • The first address then would be .217 (216+1)
  • The broadcast address would be .223 (1 less than the next multiple of 8 224-1)
  • The last address would be .222 (223-1)

Or expressed fully:

  • Subnet:
  • 1st Address:
  • Last Address:
  • Broadcast:
  • Next Subnet: (I list this because I actually figure this out then work backwards for the broadcast and last address)

This one is a bit harder – the interesting octet is the 3rd octet
Magic number=256-224=32
The interesting octet is the 3rd octet, so I focus on 178.
Recalling my 32x tables around 178

  • I see 160 is the closest multiple of 32 that is not greater than 178
  • This makes my subnet
  • The first address
  • The NEXT subnet (’cause I’m going to work backwards) is
  • So the broadcast address is
  • And the last address 1 less than the broadcast address:

Remember there are only 7 possible magic numbers (8 if you count 256-255=1), so they are not hard to learn.  Here they are
Mask Magic number
255    1 (just for completeness)
254    2
252    4 (this is a common one)
248    8
240   16
224   32
192   64
128  128

One more tricky one to finish:
Magic number=256-252=4
The interesting octet is the 3rd octet, so I focus on 56.
Recalling my 4x tables around 56

  • I can see that my subnet number must be 56 – the closest multiple of 4 to 56 that is not greater than56.  See why it is tricky?  The “interesting octet” value is a multiple of the magic number, so you stop there.
  • This makes my subnet
  • The first address
  • The NEXT subnet (’cause I’m going to work backwards) is
  • So the broadcast address is
  • And the last address 1 less than the broadcast address:

Finally, if you want to learn this stuff (like for CCNA exam) spend half an hour a night onhttp://www.subnettingquestions.com

Posted in GNS3 WorkBench | 2 Comments