Journal

bermovick

Oh yeah, I remember N WLLA OMNI. I have Doyle II as well. I'm just trying to go as thorough as I can and it's giving me quite the headache.

bermovick

Got through BGP - I pretty much understand all the concepts, although once I start getting deep enough, I absolutely get into something like the uncertainty principle.

Now I'm into QoS. I hate QoS. The weird thing is, googling for extra information on bits and pieces, this seems to be a fairly common attitude. I'm a bit curious about that. At least in a couple more chapters I get into frame-relay and multicast.

gorebrush

QoS get's me as well - so your not alone

bermovick

The good news is that this time, I'm getting a lot of A-HA! moments in these QoS chapters. WRED is finally making sense, and the switch dscp-maps so far.

With my new avatar, I was SO tempted to add an 'everyone!' after my 'good news', but thought it would be confusing to have that mid-sentence.

[EDIT/ADDENDUM]

I think some of the problems I had earlier though - and to a smaller extent this time, is due to the material not always being written all that well. Pages 554-556 and examples 13-5 and 13-6 are the perfect example.

13-5 shows putting cos 6 into priority queue 2 (along with cos 5, which is there by default), then 13-6 changes the thresholds of queue 1, and assigning cos 5 to threshold 2 and cos 6 to threshold 3. So far, so good yes?

Then the text below example 13-6 talks about how cos 5 will now drop at 60%. NO! Threshold for queue 2 where cos 5 is is still 100%!

bermovick

And another one. The 2nd paragraph on Page 558 (and indeed the rest of that page) says the Priority Queue (queue 1) cannot starve the other queues in either shared or shaped modes, but instead polices that queue for the sole purpose of preventing starvation of the other queues.

Then not 2 paragraphs later at the top of 559, it completely contradicts that, with this gem

When you configure the priority queue, the switch ignores any bandwidth values assigned to the priority queue in the share or shape commands.

Googling suggests page 559 is actually the correct one, and starvation can easily occur unless you are really picky about what gets classified into the priority queue.

reaper81

Enabling the priority queue can starve the other queues so one must be very careful if enabling that.

bermovick

Dealing with shaping today. By itself it's pretty easy, but then I hit a wall when it combines shaping AND queuing. Especially dealing with nested service-policies. I feel like I'm on the edge of understanding their interaction, but falling just short.

I drew up a diagram, trying to organize my thoughts on even the simplest situation: shaping en-mass, then classifying and queuing the excess. Here's what I came up with, and it seems to make sense that either the shaper either hands off excess to the queuer, or the queuer intercepts the packets en-route to the shaper's FIFO queue (it's a MITM!) and does its magic.

Similarly the shaper either asks the queuer for packets that are in the CBWFQ queues and the queuer grabs packets accordingly to hand back, or the queuer intercepts the shaper's check of its queue, providing sorted packets in reply.

Where I'm REALLY starting to struggle is trying to figure out if this is the only real scenario of combining the two. I can see scenarios where you'd queue first and then shape the queues ... but at that point the cliffs loom and the water at the bottom is deep and full of sharks.



PS: I don't think I've ever managed to properly format an image I'd planned to upload. I hope it's not too small to be readable.

bermovick

Here's a sample of queuing first - I have troubles here, for a number of reasons. Plus I'm not even sure if this is valid.

The obvious one off the top of my head is how can the 4 potential outputs meet the queuing requirements when they're blind to the queuing - especially those that aren't put into the shaping queues.

bermovick

Queuing done. Other than the previous notes regarding how queuing and policing can work together, I think I have a basic (possibly tenuous) grasp. So far policing has been easier, since the concepts are similar.

I have an issue with page 593 of the OCG though. For single-rate three-color policing, I disagree with the chart on how tokens are drained from the buckets. I might just be nitpicking or there's some ultra-fine detail that's not mentioned, but it says exceeding packets are those where Xp > Xbc but <= Xbe. Shouldn't it be Xp > Xbc but <= Xbc+Xbe?

bermovick

Whew. QoS done, but I'm DEFINITELY going to have to swing back there. The WAN chapter was ok - Frame Relay seems so easy when I read and set it up in my lab, but give me 24 hours of not doing it and I seem to forget half of it - nothing repeated labbing won't help with.

The first multicast chapter is all good -- I set multicast up at my current job (I was golden for a while after that, since nobody else had ever gotten it working, and it's not fun imaging 250 computers via unicast sessions).

I've run into a small sticking point in my head on some of the timers mentioned though. I actually just deleted the 3 paragraphs I wrote regarding it, but I may have figured it out -- for verification, do hosts have longer to reply to General Queries (Group Address 0.0.0.0 in the query, time is the "Query Reponse Interval" = 10 seconds) than they do to respond to Group-Specific Queries (time is "Last Member Query Interval" = 1 second)? Once I consider that possibility, the sections I'm reviewing make a lot more sense, but that just seems a bit odd.

bermovick

Well shoot - I may have to get additional material for multicast as well - not because I feel weak on it (I really don't), but because I don't feel the material I have is going quite deep enough. There are unresolved questions the book completely misses. Reviewing TCP/IP Vol 2 now to see if it helps.

bermovick

I EFFING HATE MULTICAST

... Or GNS3

... Or Virtualbox.

... Or VLC.

... Or something.

I have *NEVER* gotten this to work, and have (once again) wasted an entire day on it.

reaper81

Don't do multicast on GNS3 Or if you do make sure you read up about it. There are some nasty bugs. It has to do with the Ethernet driver or something. Try using 7200 instead of 3700 if you aren't already or just get a rack rental for doing multicast labs.

gorebrush

See - the post above is exactly why I've abandoned trying to do *anything* virtually now.

I went and dropped a load of money this week and bought me a rack and 14 devices. I've also managed to blag 3 2620's from work (two of these are XM's) and I've got a spare 3550 in the attic. I think someone said they were going to bring me a Frame Relay switch as well, so I'll have plenty of stuff to play with.

The virtual stuff is just not reliable enough. Is it a fault with the software? Is it a bug? Is it me? With physical hardware you know for sure.

I could have just bought rack tokens, but I'd never be able to sell those on at the end like I could with physical gear.

bermovick

Still here, still slogging along, but with much less confidence. Nothing seems exceptionally difficult, but the minutae, and if I slow down to even attempt to remember everything I end up moving so slowly that I end up having forgotten stuff more than 15 pages back.

I'm getting a lot of respect for people that somehow pass the written even - especially in the 100-200 hours typically given as the 'average' time taken. I think I've been at it for much longer than that, and am nowhere near confident enough to even book a date.

I'm in the latter half of the MPLS chapters on pass #2? 3? through this book - leaving only the ipv6 chapter remaining I think.

instant000

bermovick:

Hah.

It seems like you need a spaced repetition system. The only way you're going to remember it is if you repeat it. Then, the issue is how frequently to repeat it so you don't forget it, but not so often that you don't cover all of the material.

If you look through Zartanasaurus's CCIE:R&S thread, he used a spaced repetition system from the very beginning.

http://www.techexams.net/forums/ccie/76254-starting-ccie-written-just-had-talk-mrs-zartan.html

You can google "just had the talk with mrs" and see where it pops up in the results.

On this note, here's something on spaced repetition systems, to whet your appetite:
Spaced repetition - Wikipedia, the free encyclopedia

After you get a system set, the discipline comes in when you actually get your repetitions in

Hope this helps.

instant000

Another thing:

After such a long break from studying, you're probably rusty on a few things, and you've basically started over again in August. If you had stuck to it immediately after CCNP, it would be a different story.

I can also co-sign and say that 100-200 hours was a small number compared to how much I studied for the written. That 100-200 number is probably for people who manage their time well and have a solid system for studying. I'm not at that level yet. I had not studied a subject "large" enough before where a solid SRS was necessary in order for me to handle it all. I would do repetitions before, but it is not quite the same, when you can literally rip through the entire SWITCH lab manual in a single sitting, versus the CCIE, which has over a dozen topics that are EACH more in-depth than that single lab manual.

Ineffective studying techniques can cost you, when you begin working on large sets of information.

One caution:

Don't get too hung up on studying the memory stuff, or you'll find yourself three of four hours later still researching if people can truly have a photographic memory, or if it is a myth. What does that even matter if you don't have a photographic memory? How is this going to help you remember stuff? (Don't ask me why I say this, hah.)

gorebrush

Yes, I'm fairly sure I am going to have to repeat topics a few times.

Things like RIP/Switching/OSPF/EIGRP - I am solid on all the basics, mainly because I use them everyday or have practiced them enough (RIP falls into the "practice enough" category )

I think it just takes more and more repetition and labbing to get the knowledge sorted...

bermovick

For those of you who have gotten your numbers, or are (much) further along, I have to ask.

How do you keep your momentum going?

I'm not necessarily talking about motivation (although perhaps so?). Whenever I get close to booking the written, I think about the lab window and the enormity of it all and a little voice tells me I should perhaps aim a little lower - especially since I can't easily drop $300 for the written. The little voice (we'll call it 'doubt') suggests the time and $300 would be wasted because there's no way I can prepare/pass the lab anyway.

Both times this has happened I tend to get completely disillusioned and frustrated and end up abandoning the entire process for a few weeks/months because I really CAN'T afford to just drop $300 on something who's outcome is so uncertain, but then I find myself swinging back around and wanting to get back on the wagon because I want this.

(The only good news is is that letting the information 'settle' like that means each time I review I find myself remembering/understanding the concepts that I used to find difficult a little better)

gorebrush

Booking and passing the written does kinda push you into the pursuit a bit more. I *need* this.

RouteMyPacket

Where are you at work wise? Are you actually working with this technology day in and day out?

bermovick

Not really, unfortunately. We have 30-40 switches across 3 buildings here (3560,3750,4500). A couple dozen vlans that were being routed between one another on one of the 4500's when I got here. Did a lot of cleaning up old junk in the configs and turning off ip routing on the switches that didn't need it. Got multicast working & cobbled together a monitoring solution. Added some security to who can connect to them. Converted one of the inter-building links to L3 and got OSPF running across it. The other one will get done when the opportunity presents itself. There's a few other piddling things I could do, but it'd just be busywork.

Now I pretty much watch the monitor & find something else to do (studying when I'm in the mood). Today I'm ssh'd into my home linux box and have managed to get IOU working. I'm playing around with that at the moment and trying to come up with a study plan I'll be able to stick to.

bermovick

Booked for Tuesday.

I've been reviewing and I'm starting to feel like I'm spinning my wheels. We're fortunate to get the day of and the day before an exam off, so I have a 4 day weekend to try cramming my weaker areas and see how I do.

bermovick

And now begins the lab studying. My first challenge will be finding a proper plan of attack.

bermovick

The remainder of my books have started coming in. Once I get everything, here's my list:

Developing IP Multicast Networks, Vol 1
Internet Routing Architectures
Implementing Cisco IPV6 Networks
Routing TCP/IP Volumes 1 & 2
Cisco QoS Exam Certification Guide
The CCIE R&S Certification Guide.

From research, these looked to be the most widely-recommended set of "core" books (in addition to the ROUTE FLG/OCG and SWITCH OCG - but I don't know how useful those will be). Obviously this will change for v5 for its new material (dmvpn, etc), but I don't think any of these books will lose their value. If anyone else has other good suggestions, feel free to chime in.

Finally I'm a bit terrified reading other threads here. The details/depth just seem so far beyond any of the materials I've covered to date. I'm worried I'll feel I have things well in hand based on the material and get laughed at cause I studied (to everyone else) the equivalent of "Dick and Jane"

Edit:
Terror #2 is realizing that 18 months is only 76 weeks and one of those is already gone. 1% of my window is past and I haven't even started yet.

FloOz

Some of the CCIEv5 material is starting to slowly roll out. Though these links my be helpful to you.


Amazon.com: Cisco CCIE Routing and Switching v5.0 Configuration Practice Labs (3rd Edition) (Practical Studies) eBook: Martin Duggan: Kindle Store

CCIE Routing and Switching v5.0 Official Cert Guide Library (5th Edition): Narbik Kocharians, Peter Paluch, Terry Vinson: 9781587144929: Amazon.com: Books

bermovick

Oh yeah - I forgot about the OCG pack for V5. That's definitely in my to-get list.

FloOz

Yup I got it on preorder. I think it's so cool that Narbik is a co-author. I'm currently working through his workbooks and they have been really great. I expect the v5 OCG to be just as good.

bermovick

Shoot. I just realized I got the First Edition of TCP/IP Volume 1 (the 1998 one, not the 2005 one). That may explain why I got it for $1.

How much am I missing out from the Second Edition? That'll be another $40 to grab it and money is always an issue

[UPDATE]
Looks like the big changes are the removal of IGRP to legacy status, and adding (or increasing) IPv6 in 2nd edition.

bermovick

Sigh.

I have some pretty major holes in my book list. There's no layer 2 there, no MPLS there and no DMVPN there either.

MPLS and DMVPN books could probably wait a bit -- get strong on the core at layer 2 and 3, and while MPLS seems core-ish, I think it'd would be at the end of "core materials" or very early in "non core".

Research shows a lot of people seemed not to use any material dedicated to layer 2, which I can (somewhat) see -- a lot of layer 2 is ... I don't want to say "easier" than layer 3, but there just doesn't seem to be as much to it -- vlans, trunking, etherchannels and spanning-tree is most the layer 2 section of the lab, and how much more can there be to vlans, 802.1q, lacp/pagp and *stp? But that question goes both ways - how do you know how much more there is to know if you don't know how much more there is to know!

For now I guess I'll use Google for the few layer 2 technologies that are new-ish (vtp v3 and erspan), and keep an eye out at people posting or blogging about v5 book lists for comparison.