Alcatel-Lucent, Juniper Get Core-Router Upgrades

COMMENTS

Newest First | Oldest First | Threaded View

Page 1 / 5 > >>

Soupafly 12/5/2012 | 5:32:33 PM

re: Alcatel-Lucent, Juniper Get Core-Router Upgrades

Interesting announcement.

It seems ALU is following Huawei's IP team. They released the NE5000 platform over 2 years ago and it can be configured in a back-to-back architecture or in a multi-chassis system and act as a single virtual core PoP router.

Neither Cisco nor Juniper have anything to match it in terms of raw HP & port density. Where they have historically won is been the incumbent, marketing prowess & Carrier apathy/suspicion.

@ Craig M & Lightreading Editorial Team; In addition to port density & chipsets, we REALLY need you to start stating clearly the power consumption and rack deployment footprint. (Clearly = Calculations based on section 7 of this document; http://www.verizonnebs.com/TPR... Unfortunately, people try to bend the rules on stats & calculations, so a common baseline is critical.)

Space & power are the KEY metrics alongside port density & NSR feature-set support. These units are always in Tier 3/4 data-centre environments where the PoP fibre terminates and therefore are able to support deployment. In these environments space & power are at a significant premium!

Cisco's original CRS-1 family had a floor-loading that was astronomical and needed its own nuclear powerplant for juice, when I last consulted on a deployment a couple of years ago, as an example! Juniper were better but didnt come out smelling of roses & ALU had no play in that space, as per the article.

If Huawei is ever able to shake up its incompetent chinese IP marketing team and get its act together globally - not just deployments in a couple of networks in EMEA, AP & what was the USSR - then Cisco, Juniper and ALU are all in very serious trouble!

The NE5K may not have the feature density but it kills all of them for power, clustering capability and I believe space. Or at least it did....

Reply | Post Message | MESSAGES LIST | START A BOARD

Pete Baldwin 12/5/2012 | 5:32:28 PM

re: Alcatel-Lucent, Juniper Get Core-Router Upgrades

Soupafly - we hear you loud and clear about power efficiency.  I should be paying more attention there (although, as you point out, there are plenty of ways to bend the numbers). Thanks for calling me out on that, it's a good reminder.

AlcaLu does claim that the XRS is significantly lower-power than the competition (of course) and they apparently did a lot at the chip level to make that so. I'm at the AlcaLu launch event now and don't have time to research & process comparable numbers - betcha they're on the AlcaLu web site by now.

Reply | Post Message | MESSAGES LIST | START A BOARD

Flook 12/5/2012 | 5:32:28 PM

re: Alcatel-Lucent, Juniper Get Core-Router Upgrades

Excellent comment--yup, LR needs to look beyond just the surface numbers.

Reply | Post Message | MESSAGES LIST | START A BOARD

DCITDave 12/5/2012 | 5:32:27 PM

re: Alcatel-Lucent, Juniper Get Core-Router Upgrades

How would you suggest we "look beyond the numbers?" Buy a router and run it in our networks? I'm not sure what you're suggesting here.

Reply | Post Message | MESSAGES LIST | START A BOARD

yarn 12/5/2012 | 5:32:27 PM

re: Alcatel-Lucent, Juniper Get Core-Router Upgrades

What the surface shelf capacity numbers don't tell is that both CRS-3 and T4000 only support 16 100 Gb/s ports. Each slot equipped for 100G leaves more shelf capacity stranded. I'd also be surprised if any operator could equip more than 1 T4000 in a single rack without running into power/cooling issues.

Reply | Post Message | MESSAGES LIST | START A BOARD

Pete Baldwin 12/5/2012 | 5:32:27 PM

re: Alcatel-Lucent, Juniper Get Core-Router Upgrades

Flook - regarding the surface numbers - That was kind of on purpose this time. Both these announcements ambushed me, one while on the road, and I didn't have much time beyond typing up the basics.  I figured it was best to get at least a surface comparison out there.

Reply | Post Message | MESSAGES LIST | START A BOARD

torivar 12/5/2012 | 5:32:26 PM

re: Alcatel-Lucent, Juniper Get Core-Router Upgrades

Other vendors have looked at back to back multi-chassis before, they just chose not to do it because those who really needed non-blocking multi-chassis needed more than what two chassis could do.  

With all the talk of virtual chassis these days I think it makes sense and the fact you don't burn up any front facing interfaces to do it is great.   Hopefully it will work well from the software side of things and doing things like upgrades won't be a painful process...

In the comparison I would throw the PTX in there as well.  It's an IP/MPLS core router today, not a packet optical solution.   The 480Gbps/slot really hurts the PTX with regards to 100GE density, couldn't they get another 20G out of the slot?  :)   But at least it has split slots so you can only lose 40G and not the whole 80G if you still need 10G. 

Cisco and Juniper both passed what Huawei could do with the ASR9K and the MX960 in a smaller chassis, albeit without the multichassis capability, not to mention the PTX and whatever Cisco is working on to replace the CRS.   Was never really that impressed with the NE5K other than the back to back MC ability.   Then again I wouldn't be surprised to see Huawei come out with a next gen platform as well, the NE5K isn't exactly new. 

Reply | Post Message | MESSAGES LIST | START A BOARD

optiblues 12/5/2012 | 5:32:25 PM

re: Alcatel-Lucent, Juniper Get Core-Router Upgrades

You don't need to buy a router and run it in your network to go beyond the numbers. Talk to the potential end users of the router. Talk to Comcast, TWC, Verizon, AT&T, etc as a followup... see what challenges they have in deploying such a router. Floor loading, cooling,  DWDM transport and stranded capacity all have a significant impact on the gap between what is stated in a press release and what is actually deployed. I'm not picking on ALU. It is the same for all the vendors.

In general heat dictates everything we do in enigneering.

Congratulations to ALU for pushing the envelope.

Reply | Post Message | MESSAGES LIST | START A BOARD

Flook 12/5/2012 | 5:32:25 PM

re: Alcatel-Lucent, Juniper Get Core-Router Upgrades

Craig: That was definitely not a slam againt your article--and I do appreciate the time pressures of deadlines. My post was in response to what (IMO) was an excellent post by Soupafly. I guess my comment was a bit off the cuff...what I meant was that, in a comparison such as this, it would be great (assuming time and circmstance permitted) to take a bit deeper look (e.g. power/cooling) than just capacity/ports etc.

 

So no insult/criticism was intended, and apologies extended if you saw it that way. Guess I too was under time pressure of multitasking (only have one head/brain) when I shot off that post...

 

Reply | Post Message | MESSAGES LIST | START A BOARD

uguess 12/5/2012 | 5:32:24 PM

re: Alcatel-Lucent, Juniper Get Core-Router Upgrades

First, congras to the ALU IPD engineering team.  It is a great accomplishment.

But some of the marketing claims seem hard to find any support data:

 - 5X capacity.  Seems not true.  400G * 20  = 8T/system.  Existing platforms from other vendors have more than 8/5 = 1.6T.  Like others said, Cisco CRS-3 has 140G*16 = 2.24T, and Huawei NE5000E has 200G * 16 = 3.2T.

- Only and first network processor based core router.  Seems not true.  I believe Cisco CRS-3 has programmable forwarding ASIC. 

- 66% less power.  Not sure which one was the XRS compared to.  Could not find any power numbers in the datasheet.

BTW, it seems not clear whether the 400G line-card is ready to ship or not.

 

It is obviously great to have more capacity with more capabilities, however I am not convinced that changing a simple core to a complex edge is a good idea.  Not to mention that their NP might not be able to keep up with the bandwidth growth into the terabit era without reducing complex edge capabilities.

Reply | Post Message | MESSAGES LIST | START A BOARD

Page 1 / 5 > >>