sk1939Premium
join:2010-10-23
Washington, DC
kudos:9 Reviews:
 ·T-Mobile US
| reply to tubbynet
Re: NX-OS said by tubbynet:said by sk1939:Any suggestions as to where to rent rack time?
i doubt that you'll find it at this point. ccie d/c written is currently in beta. once it is moved out of beta, i'm sure you'll see places like ine and ipexpert pop out with ccie d/c vrack rentals (though i'm sure you'll pay a hefty price). either that or you can con your way into a customer site that currently has them installed. q. Crap, considering it will probably be too late (taking the beta in the next month).
Your company hiring tubby? :P |
|
 tubbynetreminds me of the danse russePremium,MVM
join:2008-01-16
Chandler, AZ
kudos:1 | said by sk1939:Your company hiring tubby? :P
always hiring.
just depends on what you're good at and whats open.
traveling is pretty much a must if you're on the post-sales side.
q.
--
"...if I in my north room dance naked, grotesquely before my mirror waving my shirt round my head and singing softly to myself..." |
|
sk1939Premium
join:2010-10-23
Washington, DC
kudos:9 | Cisco gear, management practices, dealing with clients.
Not a problem, although relocation may be...difficult. Really won't consider anything until next year though, still have to get through the written first of course. |
|
| I'm actually liking NX-OS. We're running 2 7k's for our DC switching and 2 5k's which aggregate our dozen or so 2k's in a top of rack solution. Still using 6509's in a VSS pair for redundancy which handles the core switching.
NX-OS though is pretty nice. Some of the new features come in handy and others just take some to get used to.
--
"There are two American flags flying on the property I reside on. Anyone who tries to take them down will be rendered inoperative." -Lindy |
|
tubbynetreminds me of the danse russePremium,MVM
join:2008-01-16
Chandler, AZ
kudos:1 | said by phantasm11b:I'm actually liking NX-OS. We're running 2 7k's for our DC switching and 2 5k's which aggregate our dozen or so 2k's in a top of rack solution.
good use case for the n5k. you running 2248 or 2232?
Still using 6509's in a VSS pair for redundancy which handles the core switching.
NX-OS though is pretty nice. Some of the new features come in handy and others just take some to get used to.
q.
--
"...if I in my north room dance naked, grotesquely before my mirror waving my shirt round my head and singing softly to myself..." |
|
sk1939Premium
join:2010-10-23
Washington, DC
kudos:9 Reviews:
·T-Mobile US
| said by tubbynet:said by phantasm11b:Still using 6509's in a VSS pair for redundancy which handles the core switching.
those could also easily be replaced with n7k. you'll thank yourself later. I don't know, depending on how much data is flowing between the datacentre and the core, combining the two might be more trouble than it's worth. |
|
tubbynetreminds me of the danse russePremium,MVM
join:2008-01-16
Chandler, AZ
kudos:1 | said by sk1939:I don't know, depending on how much data is flowing between the datacentre and the core, combining the two might be more trouble than it's worth.
how so?
n*10gbe between aggregation and core is the same whether or not you're sitting on two pairs of physical boxen (n7k to vss) or via vdc (n7k in vdc). the only issue comes into routing table size, tcam carving, memory carving, etc on the linecards (as you're pulling resources on a l/c between contexts). however -- in small-to-medium d/c -- this is no issue. for large d/c -- you can just slap in a pair of 7009 or similar and get rid of the c6k (or go balls-out and drop in some asr9006 or 9010 and get some real routing in the core).
q.
--
"...if I in my north room dance naked, grotesquely before my mirror waving my shirt round my head and singing softly to myself..." |
|
|
|
| reply to sk1939
said by tubbynet:I'm actually liking NX-OS. We're running 2 7k's for our DC switching and 2 5k's which aggregate our dozen or so 2k's in a top of rack solution.
good use case for the n5k. you running 2248 or 2232? 2248's. They work fine for what need them for. Our 10GE stuff for the SAN and NAS is handled by the MDS line of switches. Overall it's a good fit.
We could offload the VSS pair, in fact we've discussed it. But right now we're pretty set on VSS pairs for core switching and routing and the 7k's for DC switching. I've worked with the ASR's and have them to be buggy. A lot of HP though.
--
"There are two American flags flying on the property I reside on. Anyone who tries to take them down will be rendered inoperative." -Lindy |
|
tubbynetreminds me of the danse russePremium,MVM
join:2008-01-16
Chandler, AZ
kudos:1 | said by [phantasm11b :I've worked with the ASR's and have them to be buggy.
which line?
asr1k != asr9k != asr901 != asr903
just because it carries the prefix -- doesn't make it the same.
q.
--
"...if I in my north room dance naked, grotesquely before my mirror waving my shirt round my head and singing softly to myself..." |
|
| said by tubbynet:said by [phantasm11b :I've worked with the ASR's and have them to be buggy.
which line? asr1k != asr9k != asr901 != asr903 just because it carries the prefix -- doesn't make it the same. q. 1k's. We had issues with them for a bit and it took Cisco awhile to figure out our issue. I honestly don't remember the problem in detail but due to it the devices sat racked for a year without being used. It took a few code upgrades to get them stable.
--
"There are two American flags flying on the property I reside on. Anyone who tries to take them down will be rendered inoperative." -Lindy |
|
sk1939Premium
join:2010-10-23
Washington, DC
kudos:9 | reply to phantasm11b
In reality you could also hand off FC and FCoE to the Nexus devices if you were so inclined, but since you already have the MDS, why not use them. |
|
sk1939Premium
join:2010-10-23
Washington, DC
kudos:9 Reviews:
·T-Mobile US
| reply to tubbynet
said by tubbynet:said by sk1939:I don't know, depending on how much data is flowing between the datacentre and the core, combining the two might be more trouble than it's worth.
how so? n*10gbe between aggregation and core is the same whether or not you're sitting on two pairs of physical boxen (n7k to vss) or via vdc (n7k in vdc). the only issue comes into routing table size, tcam carving, memory carving, etc on the linecards (as you're pulling resources on a l/c between contexts). however -- in small-to-medium d/c -- this is no issue. for large d/c -- you can just slap in a pair of 7009 or similar and get rid of the c6k (or go balls-out and drop in some asr9006 or 9010 and get some real routing in the core). q. Perhaps, but your talking about buying a lot of line cards depending on the number of distribution devices you have to aggregate. I know in our case we have about 96 devices that have to be dual homed at the core, and with the ASR's that can get expensive. |
|
tubbynetreminds me of the danse russePremium,MVM
join:2008-01-16
Chandler, AZ
kudos:1 | reply to phantasm11b
said by phantasm11b:We had issues with them for a bit and it took Cisco awhile to figure out our issue.
the 1ks were a little odd in their first appearance -- but it was more software issues. the hardware is very top-notch and the platform is maturing rapidly. they have turned out to be nifty little boxen that are prime for the enterprise edge/small pop role. especially since things like vasi interfaces are (finally) supported well, otv is coming to the platform, and high-speed security features are gaining traction -- the box is sweet. add that in with the control-plane scalability on the rp2 -- and you have a box that is just as happy doing line-rate marking/nat as it is eating vpnv4 routes as an rr.
asr9k's are serious iron. iox is the bees knees.
q.
--
"...if I in my north room dance naked, grotesquely before my mirror waving my shirt round my head and singing softly to myself..." |
|
tubbynetreminds me of the danse russePremium,MVM
join:2008-01-16
Chandler, AZ
kudos:1 | reply to sk1939
said by sk1939:Perhaps, but your talking about buying a lot of line cards depending on the number of distribution devices you have to aggregate. I know in our case we have about 96 devices that have to be dual homed at the core, and with the ASR's that can get expensive.
dual-homing at the *collapsed* core is different than dual-homing at the *core*.
if you're looking for big dumb packet pushers, you can't get much sweeter than the n7k.
run some f248 l/c's on fab2 -- you'll get pure line rate forwarding on all 10gbe interfaces on all cards. it'll eat packets at layer2/layer3 all day and never flinch.
have those agg's tie into a real core (asr9k) and perform any queuing/shaping/policing to wan edge block and you've got some flexibility to make things happen.
q.
--
"...if I in my north room dance naked, grotesquely before my mirror waving my shirt round my head and singing softly to myself..." |
|
sk1939Premium
join:2010-10-23
Washington, DC
kudos:9 Reviews:
·T-Mobile US
| True, still requires purchase of a ASR9k or 1k at the very least which can get pricery depending on organizational budget. ($300k for the ASR9k).
Also true, but your still looking at $50k per 48 Port 1G blade (Bogota), and another $50k for the 7010Xl License or $70k per blade for the 32 Port 10Gig module, plus license, plus optics cost. It depends on budget and need I guess.
Ideal setup, but very very costly. |
|
tubbynetreminds me of the danse russePremium,MVM
join:2008-01-16
Chandler, AZ
kudos:1 | said by sk1939:Ideal setup, but very very costly.
(a) you can't always look at list  . if you're going to be dropping some serious coin on a d/c retrofit that involves n5k/n7k/asr9k -- you're going to have some bargaining power with your cisco account team/var account team. in all honesty -- while the numbers are harder to swallow -- you're *going* to get a better deal with a large purchase, both on hardware and support unless you're a large customer who does a lot of everyday run-rate business with cisco.
(b) it sounds trite -- but things that are bigger, better, and more flexible are generally more expensive. n7k is meant for large amounts of d/c switching/routing traffic and creates a larger level of enablement over the c6k. the c6k is cheaper. will it work? yes. are you going to be pushing the limits of the platform sooner than on the n7k? yes, especially when you consider the ability to perform migrations from sup1 to (future) sup2, fab1 to fab2, and cisco works with customers on "brownfield" migrations of linecards (i.e. m1 --> f2).
c6k will work as a "core router", but does it have (severe) limitations? yes. the asr9k provides larger 10gbe density, growth to 40gbe and 100gbe linecards, and a larger software featureset by migrating to iox. will it cost more? yes.
it depends on where the priorities and leverage lies. expensive, yes. but it gives you *a lot* that will grow with you.
--
"...if I in my north room dance naked, grotesquely before my mirror waving my shirt round my head and singing softly to myself..." |
|
sk1939Premium
join:2010-10-23
Washington, DC
kudos:9 Reviews:
·T-Mobile US
| True, but unless your doing an initial upgrade to n7k, your going to be purchasing piecemeal. I worked with a organizations that are struggling to find the money for upgrades to 10 Gig ($27k for a 6-port blade for the C6k), much less to the Nexus platform. Also true, but this is working off of the consideration that the organization is hitting the limits of the C6k enhanced platform, which while a few are, many aren't. Again true, but 40Gb and 100Gb is not seeing as wide-spread adoption as of yet with most organizations. Also, this is working off of the assumption that all of this is to be purchased. A lof of the C6k installations are existing, the number of initial purchases are dwindling rapidly, mostly in favor of either the Nexus line for Core/DC or for Juniper's offerings.
Priorities indeed, although most of my clients would balk when I suggest upgrading to a N5k in the datacenter because of the cost, much less an N7k. Remember that most organizations are working off of the concept of "good enough" or enough capacity for the next year or two, and not worrying about a few years down the road (educational institutions are a great example of this, we had a client still running 2500 series routers!). |
|
| said by sk1939:we had a client still running 2500 series routers!). *Barf*
...Dare I ask what kind of circuit(s) was hooked up to them?
Regards |
|
sk1939Premium
join:2010-10-23
Washington, DC
kudos:9 Reviews:
·T-Mobile US
| said by HELLFIRE:said by sk1939:we had a client still running 2500 series routers!). *Barf*...Dare I ask what kind of circuit(s) was hooked up to them? Regards Originally T1, then bonded T1. The ish really hit the fan when they tried to move to cable a year or two back (2009/2010). They complained they were not getting the bandwidth they were paying for, and were constantly going down (which is when we were called in, was a different company before). |
|
cramer
join:2007-04-10
Raleigh, NC
kudos:7 | Hah. T1, ok. 2 T1's, ok. 10+Mbps cable... umm, just no. (the internal forwarding rate is only 10M TOTAL.) 1600 and 1700 also have the same limits -- WIC module throughput is limited to 10M.
I still use a 2610xm on a Megapath T1. I retired my 1760 after TWC increased speeds to 10/1 -- it couldn't keep up. (3745 now. if I didn't use it for VoIP, I'd use a pix, asa, or PC.) |
|
