Saturday, October 16, 2010

CUCM 8.0(2c) on VMWare

After much trial and error, I was able to get CUCM 8.0(2c) installed on VMWare Server. In spite of only being officially supported on ESXi, it is possible to get CUCM 8 running on VMWare Server or Workstation. The best procedure to doing this is:

1. Download OVA file from Cisco and open with Text File to see recommended settings. Some of the hardware specs in the OVA file are required (e.g., dual 80 GB Hard Drives), but others are not strictly checked for during installation, like 2304 MB of RAM (2048 MB worked for me).

2. Configure a new VM with VMWare server using Red Hat Enterprise Linux 4. Any other flavor of Red Hat will not work. The following VM settings worked for me:


  • RHEL 4
  • 1 CPU
  • Dual 80 GB SCSI Hard Drives (0:0 and 0:1)
  • 2048 MB of RAM
  • USB Controller
  • IDE CD ROM
  • Ethernet Bridged
  • No need to pre-allocate the drives (you can let them grow as needed, though performance will suffer slightly). You can pre-allocate the drives if you want, though. It just takes much longer to set up the VM and takes away a lot of disk space unnecessarily.
You'll then get the Demo license which is good for a single node and 50 DLUs.

Aaron

Wednesday, July 28, 2010

CUCM 7 Features

Cisco has been hard-at-work adding new features to CUCM.  Here's a spreadsheet I've created with descriptions of all the new features for CUCM 7.  I'll soon be publishing sheets for CUCM 7.1 and CUCM 8.  Enjoy:

https://spreadsheets.google.com/pub?key=0AjU-M37pyCMfdEFtYUlWUFZlYmF3bk5NZ1lFbzFaWkE&hl=en&single=true&gid=0&output=html

Saturday, July 17, 2010

CDP Holdtime and LabUpdate

You may have shut down a link or suspected a link failure but seen "sh cdp neigh" still displaying an active connection, well that is all due to the CDP Holdtime.  It has to run down to 0 before the neighbor is removed from the table.  The default is 180 seconds, which can certainly lead to some confusion if you are troubleshooting a recently failed link.  The default can be changed with the "cdp holdtime" global parameter.  Once you're used to looking for it, you can see that the holdtime is unusually low for the failed link, which should lead you to figure it really is down w/out waiting for the entire holdtime to expire:

Rack1SW2#scn
Capability Codes: R - Router, T - Trans Bridge, B - Source Route Bridge
S - Switch, H - Host, I - IGMP, r - Repeater

DeviceID   Local Intrfce Holdtme Capability      Platform      Port ID
BB2            Fas 1/0           125          R S I            3725           Fas 0/0
Rack1SW1 Fas 1/9            15            R S I            3725           Fas 1/9

Notice that the Holdtime will keep decrementing and be unusually low (15 here) which indicates you may have a problem before you'll actually see the neighbor drop off completely.

As for my R&S lab, b/c I've just taken on a new large project I'm going to have to move my lab back til next Spring.  I'm just able to work through about 10 pages / day of INE Workbook 1, so it'll be a while before I finish!  Once I've finished WB 1, I'll then do full labs for a couple of months before attempting the real thing.

Saturday, July 10, 2010

Real Switches vs Dynamips for switching

I've gone back-and-forth with whether or not to purchase real 3550 and 3560 switches for my home lab.  I've decided NOT to, and to continue for now with Etherswitch modules.  My reasoning for this is that I prefer renting racks for 8 hours to do real labs anyways, which allows me to switch up the rack configs and to force me to sit there and run a full lab non-stop.  So, since I'll be renting racks anyways, I'll just rent some time to practice switching stuff as well.  I also don't feel like spending so much extra time getting good deals on all the parts and re-configuring the lab.

Here's a list of the major features I'll be lacking with Dynamips, and they are significant:
  • No ISL trunking
  • No bpdu-guard on interfaces
  • No QinQ trunking
  • No private VLANs
  • No PAgP / LACP
  • Cannot set switch ports to nonegotiate
  • lose some STP features, like bpduguard, bpdufilter
  • No MSTP
  • No layer 3 port-channel 
  • No Switch Database Templates
  • No udld
  • No port security
  • No port protected
  • lose some qos specific configuration of hardware queues like srr-queue
  • No extended VLANs, unless transparent mode
  • Can't modify prune-eligibility of ports

I'll still prefer to rent racks to get practice.  Luckily, http://www.gradedlabs.com/ is pretty cheap!


Friday, July 9, 2010

Whew! Getting the hang of this.

As I was planning all-along to do a hybrid of GNS3 and Lab Rentals, I went ahead and bought the standard INE workbooks (not the Dynamips version).  Well, there has been a bit of a learning curve with that.  I had to learn about re-mapping the ports b/c of only 16 ports on the ESW modules (as opposed to the 24 on the real 3560/3550s) and configuring all the router ports as fixed 100/FULL.  Then, I had to learn to THROW AWAY the Phy Diagram, and rely on the Logical and "Sh CDP Neigh" to get the configs done!  That, and not to take all of the solutions for INE Vol I too literally b/c there are some errors in the tasks and the order of the tasks may not match the verification.  That only improves your troubleshooting skills though!

If you run into any of the same issues, there's plenty of discussion on the INE forum, or feel free to drop me a line.  I'm once again re-assured that the INE materials will be fine...in particular b/c of the support from the instructors and forums.

I'm going to come up with a realistic schedule to finish Vol I once I get rolling and see how much I can reasonably cover each week.

Wednesday, July 7, 2010

Familiarize yourself with v4 Lab Interface

The v4 Lab Interface is way different than in past times.  In summary, here are some interesting changes:

1. New Lab Delivery System.
2. NO PAPER Lab Workbook.  Everything is now on Lab Delivery System app.
3. Uses Putty WITHOUT Tabbed sessions.
4. It is a little kludgy, so get used to it!

Don't waste a lab attempt not being familiar with these important changes!

Here's a good video demo:

https://learningnetwork.cisco.com/static/video-wp/CCIE_RS_v4_Lab_Exam_Demo_640x480-v2.htm

Monday, July 5, 2010

GNS3 Up and Running

I recently spent a day getting my new server (AthlonII-Quad, 4 Gig RAM, Windows 7 64-bit) up and running with the INE topology.   I ran into a few snags along the way:

1. UltraVNC is the best VNC App with Windows 7, as it is a 64-bit app and runs fast.  RealVNC and TightVNC are still 32-bit and are awefully slow.  A few notes aboue UltraVNC:

  • Be sure to set a different password for the view-only Password than you do for the VNC Password
  • In the admin properties "disable viewers inputs" must be disabled - I had to uncheck before mouse inputs would work it worked.
2. With GNS3, I had to split the INE topology into two hypervisors to get it to work. Otherwise I could only emulate 10 routers are a time!  Thanks to Mady Cisse over on the INE boards with the assist:

http://ieoc.com/forums/p/8504/106534.aspx

My settings of interest were:

  • mmap = False
  • idlepc = 0x60c05948
  • IOS 12.4-15-.t11
With this setup, my CPU is running around 35%.  Now, time to do some real work!

Thursday, July 1, 2010

CCIE R&S Preparation has Begun!

Well, I've finally finished reading the glorious 1000-page CCIE Routing and Switching Certification Guide v4 by Wendell Odom, Rus Healy, and Denise Donohue, and I've passed the Written Exam; so, it's time to begin lab prep! That Odom book was really great, and I really enjoyed the IPv6 and MPLS chapters.  Most of the errors I found were corrected in released Errata; however, one still has not. Don't let this trip you up: regarding 3560 Switch QoS on page 558 it states the following in regards to Egress Queues:

"Both shared and shaped mode schedulers prevent the PQ from exceeding its configured bandwidth when all the other queues have frames waiting to be sent. The shaped scheduler never allows any queue, PQ or non-PQ, to exceed its configured percentage of link bandwidth, even if that means that link sits idle."


"The expedite queue is a priority queue, and it is serviced until empty before the other queues are serviced."

I ran this by a few other engineers, and they also agree with this assessment, that it will indeed starve the other queues unless you explicitly configure policing.  I'm also surprised that the book never mentions that Spanning-Tree BPDUs are sent to the MCAST address 01:80:C2:00:00:00.  There's also not much detail on QinQ tunneling, and I had to access other resources to get the config portion.

When I was preparing for the the Voice lab, I spent about 18 months! That included about 8 months to complete the CCVP, 3 months to prepare for the Written, and about 5 months of prep for the Lab. And now another journey is under way. I tentatively have a lab scheduled for Oct 2010, and I plan on going through all four INE books before then. I was shocked however to recently open up the first workbook and find 2600 pages waiting to be covered! It's gonna be a busy 6 months ahead!