Random Thoughts

Read­ing Time: 4 min­utes

Random (and not so deep) Thoughts by Some Clown

I haven’t been writ­ing a lot late­ly, most­ly due to a com­bi­na­tion of my work and study sched­ule.  I thought, how­ev­er, that it would be use­ful to just toss down a few ran­dom thoughts on the prover­bial paper to wrap up 2010.  I’ll try to keep it some­what cohe­sive, but I can’t real­ly guar­an­tee any­thing.


Hav­ing made the deci­sion last year at Cis­co Live to final­ly buck­le down and pur­sue the CCIE Rout­ing and Switch­ing cer­ti­fi­ca­tion, I have been as busy as you might imag­ine with study­ing.  As I’ve gone down this road I’ve noticed a cou­ple of things:

(1) In the office I’m used to study­ing large white papers, doc­u­ments, man­u­als, com­mand ref­er­ences, etc., quick­ly to get to the answers I need for either deploy­ment or break-fix.  This is not the best way to study for the CCIE qual­i­fi­ca­tion exam, how­ev­er, as I tend to just as quick­ly for­get that infor­ma­tion past the point of it being imme­di­ate­ly use­ful.  I’ve had to change my habits now to include tak­ing notes, review­ing por­tions over and over, and cross-ref­er­enc­ing with mul­ti­ple sources.  Noth­ing earth shat­ter­ing to be sure, but a change for me.

(2) As allud­ed to above, I do a lot of cross-ref­er­enc­ing on my study mate­r­i­al.  I have mate­r­i­al from CCBoot­Camp that I con­sid­er to be my pri­ma­ry source (by virtue of being enrolled in the Cis­co 360 pro­gram through them).  I have also been read­ing the CCIE Rout­ing and Switch­ing Cer­ti­fi­ca­tion Guide, 4th Edi­tion, as well as the CCIE Rout­ing and Switch­ing Exam Quick Ref­er­ence Sheets–both by Cis­co Press.  I think it helps me quite a bit to read dif­fer­ent per­spec­tives on the same mate­r­i­al; to see it put a dif­fer­ent way on the page.  I have a Cis­co Live Vir­tu­al account as well, and so have been pulling some presentations–notably on QoS–from that site.

(3) I have over 16 years of pro­fes­sion­al expe­ri­ence in this indus­try, and while I am by no means an expert, I am con­fi­dent in things that I know.  To that end I would say that at some point in your stud­ies you will be almost guar­an­teed to come across infor­ma­tion, answers to prac­tice ques­tions, etc., that you just know are wrong.  I’ve had to learn not to be afraid to chal­lenge my study mate­r­i­al.  I don’t do it blind­ly, but I do go out and research in oth­er sources to ver­i­fy what I think I know.  I have found many instances of incor­rect infor­ma­tion in sev­er­al sources–more often than not in the Cis­co IOS exam­ple con­fig­u­ra­tions.  Some­times using com­mands that won’t work on that plat­form, oth­er times ref­er­enc­ing non-exis­tent class-maps or access-con­trol-lists.  Less often have I found bla­tant­ly incor­rect expla­na­tions of how a thing works, but even there I have found a cou­ple of exam­ples.  I take this as a good sign, actu­al­ly; it’s a sign that I am becom­ing more aware of the details of what I am study­ing.

Interesting Design Decisions

It always fas­ci­nates and bewil­ders me to see some of the design deci­sions that oth­er engi­neers make when putting togeth­er a net­work.  Much of what we do is sub­jec­tive, and even the most expe­ri­enced experts dis­agree on a good many things.  With that said, cer­tain things just don’t strike me as par­tic­u­lar­ly use­ful and it’s my pre­rog­a­tive to com­plain about them.  My top com­plaints from recent expe­ri­ence, in no par­tic­u­lar order are:

(1) My pre­de­ces­sor who built our main dat­a­cen­ter using 4503 switch­es exclu­sive­ly: access, dis­tri­b­u­tion and core (most­ly, but we do use a col­lapsed core mod­el).  The 4500 series is great but my gen­er­al argu­ment is that they’re under-pow­ered, or at least under-fea­tured for the core (Sup II-plus) and just a bit over­pow­ered for the access lay­er.  We use PoE 1‑Gig to every port in the build­ing, but the access lay­er is still bare­ly run­ning (less than 1 per­cent uti­liza­tion ever, on any met­ric).  I think some­one got a deal or some­thing.  We’re now replac­ing the core with a pair of 6506, 720 super­vi­sor, 10-gig, etc.

(2) A main dis­tri­b­u­tion point had a sin­gle 3845 with a 100-meg Inter­net con­nec­tion, and two full DS3 links.  Con­sid­er­ing the 3845 max­es out at 45 Meg of through­put, this seems a par­tic­u­lar­ly egre­gious vio­la­tion in my mind.  We’ve now moved that to a 3945, which if under full load is prob­a­bly still a tad over-sub­scribed, but much bet­ter and the price was right.

(3) Who was it at Cis­co that decid­ed that the ASA-5510 would only have two Gig links avail­able, and only with the right license?  Why only two?  Why not three or all five?  This might be a back­plane issue, I don’t know, but it just both­ers me.

(4) My own stu­pid­i­ty in set­ting up the afore­men­tioned ASA-5510 pair (failover) with the inside and out­side inter­faces on the gig links, when I should have had the two trunk links that han­dle much more traf­fic on those inter­faces.  This will be changed soon, but I should have done it right the first time.

In Conclusion

2010 has been a good year over­all, with a lot of inter­est­ing projects, expe­ri­ences, and sol­id learn­ing had by all–or at least me.  I’m look­ing for­ward to 2011 and all of the con­tin­ued suc­cess­es and expe­ri­ences to come.  I’d also like to give a spe­cial shout-out to all of my Twit­ter col­leagues, friends, fol­low­ers, and var­i­ous clingers-on and lurk­ers.  I have found the Twit­ter com­mu­ni­ty to be an invalu­able source of sup­port, wis­dom, and occa­sion­al­ly respite from the rig­ors of the dai­ly grind.  If you’re not on Twit­ter, I’d high­ly encour­age you to give it a look.

Hap­py New Year every­one!