GroupStudy.com GroupStudy.com - A virtual community of network engineers
 Home  BookStore  StudyNotes  Links  Archives  StudyRooms  HelpWanted  Discounts  Login
When to use BGP Was: number of CCIE [7:70151] posted 06/23/2003
[Chronological Index] [Thread Index] [Top] [Date Prev][Date Next] [Thread Prev][Thread Next]


At 09:34 AM 6/23/2003 -0500, MADMAN wrote:


>Zsombor Papp wrote:
>>At 08:26 PM 6/20/2003 +0000, MADMAN wrote:
>>
>>>Mark E. Hayes wrote:
>>> > NOT being a wise-a$$ here... When is it appropriate to run BGP? I set
it
>>> > up at the last job I had because I felt it was the best way to get
>>> > redundancy for web services. I had two T-1's, ASN, and had to guarantee
>>> > 100% uptime for one of our clients. Plus the enterprise was becoming
>>> > more web dependent with services we were offering.
>>> >
>>> > Thanks,
>>> >
>>> > Mark
>>>
>>>    Were the two T1's terminating at two differant ISP's?  If so BGP
>>>would be appropriate.  If you have 2 T1's terminating at a single ISP in
>>>the same POP then no.
>>
>>What would you do if they had been terminating at a single ISP in the 
>>same POP? Or did you mean "same router"?
>
>   Most likely simple default routes.

You mean default static routes? From a pure theoretical point of view, that 
seems a bit dangerous to me (to have two default static route pointing to 
two different routers). If one of the ISP routers is not directly connected 
to the customer's router (because for example there is an Ethernet switch 
in the middle), then the customer's router won't notice even if the ISP 
router is powered down. Even if the two routers are directly connected, it 
is possible (even though with current IOS versions it is not very likely) 
that the interface stays up even though routing dies on the ISP router. In 
short, I think there is a chance of you blackholing half of your traffic.

Comments? :)

Thanks,

Zsombor

>  Though as Howard alluded to there are some exceptions where you may use 
> BGP but not to recieve full routing but more likely to control network 
> announcements.  You could exchange same router/POP as far as routing is 
> concerned.
>
>   Dave
>
>>Thanks,
>>Zsombor
>>
>>>   Dave
>>>
>>> >
>>> >
>>> >
>>> > -----Original Message-----
>>> > From: nobody@xxxxxxxxxxxxxx [mailto:nobody@xxxxxxxxxxxxxx]On Behalf Of
>>> > MADMAN
>>> > Sent: Thursday, June 19, 2003 11:59 AM
>>> > To: cisco@xxxxxxxxxxxxxx
>>> > Subject: Re: number of CCIE [7:70151]
>>> >
>>> >
>>> > n  The same was true of my 2-day
>>> >
>>> >>test, again, I had done everything on both days by mid-afternoon and I
>>> >
>>> > just
>>> >
>>> >>sat around with nothing to do but check my work over and over again.
>>> >
>>> >
>>> >    Hmm, when I took the lab you were done configuring at noon on the
>>> > second day at which time the liberty was taken to destroy what you had
>>> > built and you then had a couple of hours to put it back together.
>>> >
>>> >    Dave
>>> >
>>> >    Nor is
>>> >
>>> >>my experience unique - I think that most CCIE's would agree that if
>>> >
>>> > you're
>>> >
>>> >>not done with several hours to spare, you're probably not going to
>>> >
>>> > pass.  I
>>> >
>>> >>would venture that very few people that have  passed the test have
>>> >
>>> > actually
>>> >
>>> >>required all the of the testtime that was allotted to them.
>>> >>
>>> >>What seems to kill people is that they don't read the questions
>>> >
>>> > carefully
>>> > or
>>> >
>>> >>they simply don't know the material and then they consequently make
>>> >>mistakes, and then in their haste, they start working too fast thereby
>>> >>making more mistakes, etc.  But again, if you know the material and
>>> >
>>> > you're
>>> >
>>> >>careful about reading the questions, the test is really quite
>>> >
>>> > straightforward.
>>> >
>>> >>
>>> >>>This is also probably why I got some seriously mixed reviews
>>> >>>from
>>> >>>different CCIEs in terms of the difficulty of the exams (be it
>>> >>>one
>>> >>>day or two day).
>>> >>>
>>> >>>For the record, the one day exam was more suited to my style
>>> >>>than the
>>> >>>two day sounded like.  Oh well, I will never have a direct
>>> >>>comparison
>>> >>>now.
>>> >>>
>>> >>>The same was said about the two day as well in terms of speed
>>> >>>but
>>> >>>with some ancillary tricks such as the physical element, etc.
>>> >>>I
>>> >>>suppose that is good to know, but hey, nothing 5 minutes
>>> >>>couldn't
>>> >>>figure out on a web page.
>>> >>
>>> >>
>>> >>I agree that the physical element was dumb.  But the troubleshooting
>>> >
>>> > section
>>> >
>>> >>was absolutely critical, see below.
>>> >>
>>> >>
>>> >>
>>> >>>The troubleshooting element was definitely a sorely missed
>>> >>>element
>>> >>
>>> >>>from the two day lab, but trust me, with the one day it is a
>>> >>
>>> >>>dynamic
>>> >>>truobleshooting element built in.  It is VERY easy to break
>>> >>>your
>>> >>>working network while you perform the exam.
>>> >>
>>> >>
>>> >>But not realistic.  Let's face it - as a network engineer, how many
>>> >
>>> > times
>>> >
>>> >>are you really building networks from scratch vs. how many times are
>>> >
>>> > you
>>> >
>>> >>troubleshooting already-built networks?  The fact is, building
>>> >
>>> > networks
>>> > from
>>> >
>>> >>scratch is really only a minor part of the overall job, most of the
>>> >
>>> > time
>>> > you
>>> >
>>> >>are maintaining built networks.  A far more useful test would be one
>>> >
>>> > that
>>> >
>>> >>was PURE troubleshooting.  For example, you get the whole morning to
>>> >>familiarize yourself with the network, and in the afternoon, all kinds
>>> >
>>> > of
>>> >
>>> >>funky problems get injected into your network.  One serious problem
>>> >
>>> > with
>>> > the
>>> >
>>> >>present format is that you end up with guys who are really good at
>>> >>configuring stuff but not very good at troubleshooting existing
>>> >
>>> > networks.
>>> >
>>> >>
>>> >>>Unfortunately, because it is more speed driven and because the
>>> >>>content, while jam packed, is probably 'less', it also means it
>>> >>>might
>>> >>>be more prone to some form of bootcamp brain dumpage.  But this
>>> >>>is
>>> >>>not really conclusive. It might just be that, the CCIE is
>>> >>>becoming
>>> >>>"more popular" and people have recently tapped into this
>>> >>>market.  The
>>> >>>drop in Cisco gear pricing on the used market probably had a
>>> >>>LOT to
>>> >>>do with bringing down this barrier to entry.
>>> >>
>>> >>
>>> >>Well, the market for bootcamps is pretty darn good proof that it's
>>> >>conclusive.  Think of it logically - why would people be willing to
>>> >>consistently cough up thousands of dollars for bootcamps if they don't
>>> >>work?  Either all these people are all stupidly throwing their money
>>> >
>>> > away,
>>> >
>>> >>or you have to concede that bootcamps are making the test easier.  PT
>>> >>Barnum  said that while you can fool all the people some of the time
>>> >
>>> > and
>>> >
>>> >>some people all the time, you can't fool all the people all the time.
>>> >
>>> > If
>>> >
>>> >>bootcamps really had no value, it is likely that this would be common
>>> >>knowledge by now.
>>> >>
>>> >>
>>> >>
>>> >>>Regretably, it is difficult to say whether or not it is the
>>> >>>slippery
>>> >>>slope we are going up if we really believe a one day exam is
>>> >>>instantly easier than a two day and that is the reason why
>>> >>>there are
>>> >>>more CCIEs per month, or if it is because the failure rate is
>>> >>>the
>>> >>>same, and the expected value of passing CCIEs goes up due to
>>> >>>the
>>> >>>higher volume of candidates per month.
>>> >>>
>>> >>>Whether or not it is easy or not, I cannot say.  I encourage
>>> >>>any
>>> >>>CCIEs of the two day to take a one day and see how it is.  I
>>> >>>only
>>> >>>know of one who did it, and he felt it was worse than the two
>>> >>>day
>>> >>>lab.  But, like I said, different types of people, different
>>> >>>types of
>>> >>>problem solvers.  Might be easier for some.
>>> >>
>>> >>
>>> >>My opinion- it's easier.  Significantly easier.  Another guy who has
>>> >
>>> > also
>>> >
>>> >>taken both, John Kaberna, has said the same thing.
>>> >>
>>> >>
>>> >>But it's not just the 1-day vs. 2-day thing.  It's an entire suite of
>>> >>factors that together have degraded the difficulty of the cert.  The
>>> >
>>> > CCIE
>>> > is
>>> >
>>> >>suffering death by a thousand cuts, of which the format change is only
>>> >
>>> > one
>>> >
>>> >>cut (albeit a substantial one).  Like I said, the proliferation of
>>> >
>>> > bootcamps
>>> >
>>> >>and dedicated practice labs, and all these other things all take their
>>> >
>>> > toll.
>>> >
>>> >>
>>> >>
>>> >>
>>> >>>One thing is true though.  By law of numbers, even if the
>>> >>>percentage
>>> >>>rate of failure IS the same, since the NET number of CCIES
>>> >>>passing is
>>> >>>higher, by supply and demand the value of the CCIE is
>>> >>>dropping.
>>> >>>(someone else mentioned this as well).
>>> >>>
>>> >>>If the percentage of failure is even lower... then the value
>>> >>>just
>>> >>>drops exponentially.  :)
>>> >>>
>>> >>>As for having a lower CCIE number, I do not care, I do not
>>> >>>know.
>>> >>>Most of the really older CCIE numbers I know tend to be
>>> >>>mediocre with
>>> >>>the new technology and are sick of knob turning anyway
>>> >>>(although
>>> >>>some are still verry good).  The medium numbers seem to be the
>>> >>>best.
>>> >>>;)  The ones on the highest numbers end seem to be a mixed bag.
>>> >>
>>> >>
>>> >>I believe that people place far too much emphasis on knowing the new
>>> >>technology.  Hey, don't get me wrong, it's important to keep up.  But
>>> >
>>> > let's
>>> >
>>> >>not overemphasize this point too much.  For example, take the case of
>>> >
>>> > the
>>> >
>>> >>R/S CCIE which is the CCIE that is supposedly geared to
>>> >
>>> > enterprise-level
>>> >
>>> >>networking (those guys who want to do service-provider work are
>>> >
>>> > supposed to
>>> >
>>> >>be looking at the C/S CCIE).  Some people have retorted that the
>>> >
>>> > low-number
>>> >
>>> >>R/S CCIE's don't know, say, BGP, so they contend that the
>>> >
>>> > higher-number
>>> > CCIE
>>> >
>>> >>is actually more relevant and useful. But let's be honest - how many
>>> >>enterprises actually run BGP?  1% at most?  Probably more like 0.1%,
>>> >
>>> > or
>>> >
>>> >>perhaps even less?  And even those enterprises that are running BGP -
>>> >
>>> > how
>>> >
>>> >>many actually have a legitimate need to run BGP vs. how many have just
>>> >
>>> > done
>>> >
>>> >>it for stupid reasons (something that myself, Howard, and Peter van
>>> >
>>> > Oene
>>> >
>>> >>have discussed before)?  Even in those cases, how many actually have
>>> >
>>> > enough
>>> >
>>> >>BGP routers that they might actually need to run their own
>>> >>route-reflectors?  And furthermore,  I have to ask, how many
>>> >
>>> > enterprises
>>> > are
>>> >
>>> >>running BGP not because they actually need it, but because their
>>> >
>>> > network
>>> >
>>> >>engineer has decided to make things more complicated than they really
>>> >
>>> > need
>>> >
>>> >>to be because it means greater job security for himself/herself (i.e.
>>> >
>>> > "...if
>>> >
>>> >>I install BGP everywhere and I'm the only person here who actually
>>> >
>>> > knows
>>> >
>>> >>BGP, that makes it that much harder for them to lay me off...")?  How
>>> >
>>> > many
>>> >
>>> >>enterprises are like this?  I don't know the answer either, but it's
>>> >
>>> > safe
>>> > to
>>> >
>>> >>say that the number is greater than zero.
>>> >>
>>> >>Or take the case of IP multicasting.  With apologies to Howard
>>> >
>>> > Berkowitz -
>>> >
>>> >>pop quiz - name 10 popular IP multicasting applications that, right
>>> >
>>> > now,
>>> > are
>>> >
>>> >>in use in the company you work for. Can't do it, can you?  Can you
>>> >
>>> > even
>>> > name
>>> >
>>> >>one?  For most people, they can't even name a single one.  In all my
>>> >
>>> > years
>>> >
>>> >>of networking, I have not run into a single enterprise that is
>>> >
>>> > actually
>>> >
>>> >>actively using IP multicasting.  Now don't get me wrong - I know that
>>> >
>>> > there
>>> >
>>> >>are some rare cases of multicasting being used in the enterprise.  But
>>> >
>>> > the
>>> >
>>> >>key operating word there is 'rare'.  For various reasons, I believe
>>> >
>>> > anything
>>> >
>>> >>that could be done by IP multicasting could probably be done far
>>> >
>>> > easier
>>> >
>>> >>either through a broadcast network (for example, right now through my
>>> >>digital cableTV service at home I get hundreds of TV channels - and
>>> >
>>> > quite
>>> >
>>> >>frankly most of them suck -  and with compression algorithms improving
>>> >
>>> > all
>>> >
>>> >>the time, I may be getting thousands of channels in the near future)
>>> >
>>> > or
>>> >
>>> >>through an application-level proxy/cache/CDN arrangement.   But the
>>> >
>>> > point
>>> > is
>>> >
>>> >>that even the most fervent IP multicasting supporter has to concede
>>> >
>>> > that
>>> > the
>>> >
>>> >>technology hasn't exactly taken the world by storm.
>>> >>
>>> >>Therefore the argument that the newer CCIE test supposedly has more
>>> >
>>> > relevant
>>> >
>>> >>technologies really doesn't hold water.  In the case of BGP, most
>>> >>enterprises don't need it, in the case of route-reflection most
>>> >
>>> > enterprises
>>> >
>>> >>don't know it and care about it, and in the case of IP multicasting,
>>> >
>>> > most
>>> >
>>> >>enterprises don't know it, don't need it and don't care about it.  Or,
>>> >
>>> > let
>>> >
>>> >>me put it to you another way.  The newest version of the CCIE no
>>> >
>>> > longer has
>>> >
>>> >>IPX or tokenring.  Yet I think I'm on safe ground when I say there are
>>> >
>>> > far
>>> >
>>> >>more enterprises out there running tokenring and IPX than are running
>>> >
>>> > IP
>>> >
>>> >>multicasting or BGP route reflection.  Therefore, of the older or
>>> >
>>> > newer
>>> >
>>> >>CCIE, which one  is REALLY more relevant to present-day enterprise
>>> >
>>> > networks?
>>> >
>>> >>
>>> >>
>>> >>
>>> >>
>>> >>>And while someone said the "higher number ones" have "less
>>> >>>experience" that should not be true in theory since the CCIE
>>> >>>was
>>> >>>designed for people who already worked in the networking field
>>> >>>for
>>> >>>years.
>>> >>>
>>> >>>However, I will agree in practice, that does seem to happen
>>> >>>often
>>> >>>(higher numbers, less experience).
>>> >>>
>>> >>>I think as with all things in life, take the individual on a
>>> >>>case to
>>> >>>case basis.  You are going to find good and bad apples in every
>>> >>>basket.  The CCIE is still a very good certification, I do not
>>> >>>think
>>> >>>anyone is denying that.  But I do not think it is clear if it
>>> >>>is
>>> >>>blatantly easier now.
>>> >>
>>> >>
>>> >>I didn't say that it had turned into the CCNA.  But it's not the
>>> >
>>> > rockhard
>>> >
>>> >>exam that it used to be.  And that's not the fault of anybody here.
>>> >
>>> > That's
>>> >
>>> >>the fault of Cisco itself.
>>> >>
>>> >>
>>> >>
>>> >>>-Carroll Kong
>>> >>
>>>
>>>
>>>--
>>>David Madland
>>>CCIE# 2016
>>>Sr. Network Engineer
>>>Qwest Communications
>>>612-664-3367
>>>
>>>"Government can do something for the people only in proportion as it
>>>can do something to the people." -- Thomas Jefferson
>--
>David Madland
>CCIE# 2016
>Sr. Network Engineer
>Qwest Communications
>612-664-3367
>
>"Government can do something for the people only in proportion as it
>can do something to the people." -- Thomas Jefferson




Message Posted at:
http://www.groupstudy.com/form/read.php?f=7&i=71151&t=70151
--------------------------------------------------
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to abuse@xxxxxxxxxxxxxx