Marconi Adds 'c' to Its OC192

COMMENTS

Newest First | Oldest First | Threaded View

<< < Page 3 / 6 > >>

MP_UK 12/4/2012 | 9:43:02 PM

re: Marconi Adds 'c' to Its OC192 jamesbond:
Why combine routing and ATM switching in one box?
----------------

Good question. Sort of like asking: what's the difference between a router with ATM ports and an ATM switch with a routing engine? I don't really know about multiservice boxes, so I'll leave that question to someone else...

Cheers
MP

Reply | Post Message | MESSAGES LIST | START A BOARD

Belzebutt 12/4/2012 | 9:43:02 PM

re: Marconi Adds 'c' to Its OC192 Why combine routing and ATM switching in one box?

Because then you have a "multiservice" switch and these boxes sell well nowadays.

AFAIK, you do SARing when you convert your packet traffic into ATM cells (segement the packets). So if you have a multiservice switch that can accept traffic as packets (e.g. on a POS interface) and then put them onto an ATM core as cells, that is the device that will perform the SARing.

If that switch switches the packets as packets, then they must be SARed on the output ATM interface. If that interface is an OC-192, you need an OC-192 SAR because you will have packets coming out that interface at OC-192 rate and you will need to segment them into cells that fast.

I think it's also possible that if the switch switches everything as ATM cells, you could segement the packets on the input POS interface, and then just switch them across your ATM fabric, and the output OC-192 interface doesn't need to SAR. On the other hand, if you have an OC-192 POS interface on that particular box, you have packets coming in at OC-192 and they need to be SARed at OC-192 which then requires and OC-192 SAR.

So you don't really need an ATM SAR, but there are cases in which you could use one...

Reply | Post Message | MESSAGES LIST | START A BOARD

Belzebutt 12/4/2012 | 9:43:00 PM

re: Marconi Adds 'c' to Its OC192 Good question. Sort of like asking: what's the difference between a router with ATM ports and an ATM switch with a routing engine? I don't really know about multiservice boxes, so I'll leave that question to someone else...

If your box has full-featured routing and limited ATM capabilities (i.e. no PNNI or other ATM protocols), but it has ATM interfaces just to get IP traffic off of them, then it's a router.

If your box has really good ATM switching, full-featured ATM protocols, but limited routing capabilities, then it's a multiservice switch.

Usually when a box is designed it's geared to be a router or an ATM switch with multiservice. Not many boxes are designed to do both well, I can't think of any popular boxes that could be used both as a core router or as a core ATM switch.

Reply | Post Message | MESSAGES LIST | START A BOARD

broadbandboy 12/4/2012 | 9:42:52 PM

re: Marconi Adds 'c' to Its OC192 THE ATM & IP REPORT
E-Note #7
Sept. 19, 2002

FLASH----- Govt. Test Lab Turns On 1st OC-192c 10 Gbps ATM Interface Card

In a test network set up at the Naval Research Labs near Washington DC, a network consisting of the Marconi BXR with a new OC-192c ATM card was able to process five 1.6 Gbps streams of high definition video simultaneously earlier today.

The ATM & IP Report spoke with Dr. Hank Dardy, Sr. Technologist at the NRL. Dr. Dardy explained that the folks at Marconi got the card yesterday, and drove all night to get it there this morning by 8am.

Dardy: "We plugged it in, and by 10 am were running live traffic over a bunch of circuits. So in two hours we got two interfaces up and were passing traffic around the network."

Q. How do you generate enough traffic to run at this rate?

Dardy: "We took five high definition video streams and aggregated them onto the network. ThatGÇÖs 1.6 Gbps stream times five, so itGÇÖs pretty close. Its more than three OC-48s could handle, so I want it to be known that we are stressing the interface."

"The video stream is AAL1, although the card is agnostic to traffic type. The AAL type is generated by the source, and the card can pass any type of ATM traffic."

Dr. Dardy explained that in the past they have operated a visualization demonstration from SGI (Silicon Graphics) called Space to Face, which takes a view of the earth from space, and then zooms down through multiple layers of imagery in real time. ItGÇÖs quite a good consumer of bandwidth, requiring about 6 Gbps in aggregate to drive multiple projectors running video streams at about 1.5 Gbps each.

Q. Why not use IP for this application?

Dardy: "The best way is to run IP over ATM. The key for the government is we can encrypt the streams at OC-192 rates with Type-1 security. A new box coming soon from General Dynamics called the KG75a can encrypt ATM at OC-48c and OC-192c (2.5-10 Gbps). The previous model, KG75 could encrypt at OC-3 to OC-12 rates."

Q. Does it make a difference if you are trying to encrypt ATM cells or IP packets?

Dardy: "Yes, it makes a heck of a difference. This is the reason for wanting ATM in the DoD. Basically ATM has the state that enables us to do QoS end-to-end, and that state is also what keeps this crypto working at 100% effectiveness."

"In contrast, they are breaking their necks today with IP and they are only up to 100 megabits, while we are already running ATM at 10 gigabits. They are talking about developing an encryptor for gigabit
Ethernet by 2004 or 2005. But we canGÇÖt wait; we have a lot of data to move today. For me the limit is 10, and I am already asking them when they can do 40 Gbps, with a goal to get to 160 Gbps."

End.

P.S. Please send any questions or comments directly to the editor at: [email protected]

Reply | Post Message | MESSAGES LIST | START A BOARD

Belzebutt 12/4/2012 | 9:42:47 PM

re: Marconi Adds 'c' to Its OC192 That's quite interesting. How does Type-1 security differ from IPSec? There are commercial devices today that can encrypt IPSec well over 100 Mbps.

Reply | Post Message | MESSAGES LIST | START A BOARD

Belzebutt 12/4/2012 | 9:42:46 PM

re: Marconi Adds 'c' to Its OC192 At any rate (no pun), this interface is ahead of the game, as was the OC48c from Marconi. It's quite clear from my posts that I am a strong supporter of their equipment and F-PNNI implementation. Quite frankly, it's because the equipment/software has a track record of excellent perfomance and reliability.

No one can argue that they are ahead, it's undisputable that they have the only OC-192 ATM interface on the market. They have shown that ATM is going strong and can still be used in the core with huge switches. Fore has been a pioneer of ATM so there's no question that they know what they're doing.

By the way, Fore also had a hand in Nortel's early core ATM switches. Some even used F-PNNI.

Reply | Post Message | MESSAGES LIST | START A BOARD

NetworkMercenary 12/4/2012 | 9:42:46 PM

re: Marconi Adds 'c' to Its OC192 "Usually when a box is designed it's geared to be a router or an ATM switch with multiservice. Not many boxes are designed to do both well, I can't think of any popular boxes that could be used both as a core router or as a core ATM switch."

Belz-
I would agree with you to this point in date. I am extremely interested in seeing actual test results from a group like Tolly, etc. It probably is quite accurate to say that the results will be very positive. I can't imagine the NRL issuing statements without effectively (key word) testing the box/interface.

At any rate (no pun), this interface is ahead of the game, as was the OC48c from Marconi. It's quite clear from my posts that I am a strong supporter of their equipment and F-PNNI implementation. Quite frankly, it's because the equipment/software has a track record of excellent perfomance and reliability.

I've made a few statements in previous posts that I believe the BXR will start to find it's way into SP's networks...taking time, but it will. I have no doubt.

Note:
Connection-Oriented (C0) -> the past, present and future. All are trying to be CO. It's the only way to offer/collapse multiple flavors of services effectively. The key question is who's flavor of CO are SP's going to choose? Is it IPoverATM, IPoverSONET, IPoverEthernet, etc. or a lovely mix? The one's who decide that they want to play with all the services will choose IPoverATM I believe - NM

Reply | Post Message | MESSAGES LIST | START A BOARD

Belzebutt 12/4/2012 | 9:42:45 PM

re: Marconi Adds 'c' to Its OC192 (1) An MPLS standard that provides per flow explicit QoS mechanisms (because I think the packet industry is reluctantly beginning to understand that we need them, rather than diffserv-like algorithms), and

http://www.ietf.org/internet-d...

I'm not sure if this is the one but there is a draft.

(2) An MPLS switch product that supports that (non-existent) standard

Vivace and Laurel are doing this.

Reply | Post Message | MESSAGES LIST | START A BOARD

Ibeenframed 12/4/2012 | 9:42:42 PM

re: Marconi Adds 'c' to Its OC192 **How does Type-1 security differ from IPSec?**

IPSec is a family of tunneling techniques, authentication methodologies and encryptions techniques that can be put together in many different flavors. Last I checked (and it's been a while since I checked) 3DES was the highest level of encryption used by IPSec.

3DES was recently cleared for export as an encryption tool. I read that to mean that the real hyper parinoid security agencies don't feel 3DES to be secure anymore.

That leads to a conclusion that the encryption being done by the security agencies is much more complex and sophisticated than required by 3DES.

Reply | Post Message | MESSAGES LIST | START A BOARD

Ice Man 12/4/2012 | 9:42:40 PM

re: Marconi Adds 'c' to Its OC192 <<that's 100="" are="" can="" commercial="" devices="" differ="" does="" encrypt="" from="" how="" interesting.="" ipsec="" ipsec?="" mbps.="" over="" quite="" security="" that="" there="" today="" type-1="" well="">>

I'm not sure of the definition of Type-1 security, but I think those FASTLANE KG-75 encryptors can encrypt entire OC-48 and OC-192 ATM streams.

http://www.gdc4s.com/Products/...

<<dardy: "yes,="" a="" atm="" basically="" difference.="" dod.="" enables="" for="" has="" heck="" in="" is="" it="" makes="" of="" reason="" state="" that="" the="" this="" to<br="" us="" wanting="">do QoS end-to-end, and that state is also what keeps this crypto working at 100% effectiveness."
"In contrast, they are breaking their necks today with IP and they are only up to 100 megabits, while we are already running ATM at 10 gigabits. They are talking about developing an encryptor or gigabit Ethernet by 2004 or 2005. But we canGÇÖt wait; we have a lot of data to move today. For me the limit is 10, and I am already asking them when
they can do 40 Gbps, with a goal to get to 160 Gbps.">>

40 and 160 Gbps ATM! Just the kind of bandwidth needed to fight a war remotely?

A few years ago the "experts" said 10Gbps ATM couldn't be done.

~~ Ice

</dardy:></that's>

Reply | Post Message | MESSAGES LIST | START A BOARD

<< < Page 3 / 6 > >>