x
<<   <   Page 3 / 5   >   >>
wildcard 12/4/2012 | 8:26:19 PM
re: Nortel's Empty Room at the Top I suppose, you point is that carriers will allocate badnwidth in the form of independent networks/pipes on a per-customer basis. If customer is not using allocated bandwidth then so be it - is it correct ?

Thanks,

Netskeptic


Thus we have the disconnect...

Carriers can invoice.
Users think its pay as you go.
When the internet works like the telephone then you are on to something.
Physical_Layer 12/4/2012 | 8:26:18 PM
re: Nortel's Empty Room at the Top (Posted this elsewhere, looking for more comments)

Hi all,

I always find it interesting that EVERYBODY talks about Sonet vs DWDM but nobody ever gives an explanation, and I am starting to think there isn't on.

Example: this article talks about Ciena buying Cyras who makes SONET products rather than ONIS who makes DWDM products

Criticism: I believe that for the most part, SONET and DWDM are mutually exclusive. SONET is a protocol, DWDM is not. Instead, it is simply a means to create more virtual fibers.

I also see this same problem with RHK reports, and other firm's reports. What the heck is meant by SONET vs DWDM anyway? I mean I could understand it if somebody was to say Sonet UPGRADE vs DWDM UPGRADE, but that is not what is said.

Any help appreciated. Anyone else been confused by this before?

gladysnight 12/4/2012 | 8:26:18 PM
re: Nortel's Empty Room at the Top wildcard,

you make a good point.

my memories of early ATM evangelism goes like this:

1. You (or your end device as a proxy) contract with the network for the service qualities you need in order to complete your task, i.e. gauranteed bandwidth, latency, jitter, etc.

2. You pay the per bit price for the service type these qualities represent.

3. The carrier totes up your bit charges and invoices you.

So, for internet access you might have no qualities other than minimum throughput (i spose for browsing only you might be able to buy asymmetric service and the carrier could aggregate demand from different directions to oversubscribe full duplex capacity).

For voice you might specify less than 150ms latency and low jitter for business use, less strict qualities for personal use.

For different grades of video you need more and more bandwidth, less and less latency and jitter again.

What happened?
gladysnight 12/4/2012 | 8:26:18 PM
re: Nortel's Empty Room at the Top "I suppose, you point is that carriers will allocate bandwidth in the form of independent networks/pipes on a per-customer basis. If customer is not using allocated bandwidth then so be it - is it correct? - Netskeptic
==========


Yep, that pretty much says it.

netskeptic 12/4/2012 | 8:26:17 PM
re: Nortel's Empty Room at the Top > I'm not trying to be argumentative, just trying
> to keep up with you...

And I am just trying to get hold on the modern thinking in the area I am not that current with, and I am very grateful for the info everybody provides.

> But even for voice or video traffic, is it not
> up to the (Layer 3) end devices to develop the
> necessary signalling techniques and traffic
> management schemes ?

Speaking about voice I would say `no chance'. L3 is hard set into IP and IP does not have anything in it to provide for voice grade connections. So, it is no surprise that the search for this functionality is confined to L2 (MPLS, RPR, SONET, ATM etc).

Thanks,

Netskeptic

Petabit 12/4/2012 | 8:26:17 PM
re: Nortel's Empty Room at the Top SONET is a physical layer protocol. Like Ethernet.

DWDM is the art of mixing different colours together on a single fibre.

Pretty much all SONET gear these days is DWDM capable - you just need to put more than one colour of traffic on a fibre.

So you are right that 'everyone' is confused about the difference. Interestingly none of the techies try and argue SONET v. DWDM, only the analysts and trade press.

What RHK mean (even they have got the names wrong) is the difference between SONET-only DWDM equipment (closed interface) v. any-protocol DWDM equipment (open interface). Most of the metro gear will carry native ESCON, Ethernet, FDDI as well as SONET - so they get called 'DWDM'.

So they should be calling them 'open' and 'closed'. Interestingly by the RHK definition a card that takes eight GbE streams and squirts them over a 10G DWDM wavelength is a 'SONET' box since it can only transport one protocol. If that makes any sense to you, give me a call.

The point is more that closed interfaces are on the way out, most of the newer boxes will transport anything you throw at them. So the distinction is becoming a moot point.

P.

netskeptic 12/4/2012 | 8:26:16 PM
re: Nortel's Empty Room at the Top > ATM has a huge overhead (around 10%)
> Packets over SONET on the other hand have a
> much smaller overhead. (1-5% depending on
> packet size)

My guess is that the necessity to have high speed SAR processing is a real killer.

Plus core bandwidth reservation proved to be not as big an issue due to massive network build up.

However, the simple fact that ATM is alive and kicking despite all its shortcomings points to the fact that network needs hard bandwidth reservations with fixed traffic characteristics.

Just an IMHO.

Thanks,

Netskeptic




ranon 12/4/2012 | 8:26:16 PM
re: Nortel's Empty Room at the Top > What happened?

What I believe happened is this.

ATM has a huge overhead (around 10%)
Packets over SONET on the other hand have a much smaller overhead. (1-5% depending on packet size)

One advantage that ATM had over IP was that it was easier to switch based on VPI/VCI than it was to route based on the IP header.

So initially ATM was used in the core networks.
Once IP routers became faster, ATM lost that space.

Correct me if I am wrong but the overheads killed ATM.

ranon
Belzebutt 12/4/2012 | 8:26:15 PM
re: Nortel's Empty Room at the Top "ATM has a huge overhead (around 10%)"

I think it's actually a lot higher for TCP/IP traffic, especially since half of IP traffic is 64-byte packets that need to go into two cells, plus the AAL5 overhead.
cfaller 12/4/2012 | 8:26:15 PM
re: Nortel's Empty Room at the Top "What happened?"

In addition to what ranon pointed out, the cost of ATM never came down enough in order for it to become ubiquitous. A lot of times (not all the time, of course), ATM service was more expensive than private lines! That will never work, and I think the root cause was that the ATM equipment cost so darn much that carriers couldn't get a good return...
<<   <   Page 3 / 5   >   >>
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE