Several factors could come into play, including the need for that type of bandwidth, latency factors, and resiliency. If a client needs an extremely resilient connection with low latency, and high bandwidth, a protected SONET OC3 solution is the best choice, due to the nature of the architecture.
If you have strict requirements in regard to latency and line quality OC3 could be a good choice as an alternative to ethernet services.
With OC3 you can have complete end-2-end monitoring of bit errors, and it is also dedicated 100% for your usage. Ethernet services lack the full end-2-end monitoring capabillity, as frames could get dropped alog the way.
The CPE equipment will cost you more for OC3, than for ethernet. The built in redundancy in SONET/SDH can provide sub 50ms switchover easily, but to harvest the full potential of redundancy you must offcourse have redunant fiber to your location.
Conditions to consider
1) Data usage requirements
2) Redundancy is a feature, not a requirement. Therefore the architecture dictates the requirement and fees.
3) Other (possibly concatenated) options that are available to provide similar bandwidth
Optimal Conditions
1) Several key hub locations (forming the OC3 ring) that are well connected to the remaining network
2) A CONTINUOUS bandwidth requirement of more than DS3 24x7
3) A well integrated voice and data architecture that can run across this backbone and therefore reduce PSTN and Dedicated Access Data line cost.
4) Existing purchasing relationships with the CPE equipment vendors, allowing you to reduce CPE costs
I think everyone would agree on a monetary standpoint that when the costs of DS3's exceed the costs of having the OC-x circuitry then it's time to begin to think about a switch. The problem lies in the fact that you have already heavily invested in the DS3 hardware and now will have to purchase OC-x hardware to begin a migration. So you have to factor in hardware costs with the overall cost equation.
Secondly you have to keep in mind redundancy and latency. If your network is mission critical and requires no downtime then it becomes a bit more complicated. Ordering 1 OC3 on sonet still does not guarantee no downtime. To ensure that this downtime is kept to a minimum you have to order 2 OC3 circuits. When ordering the first one you want it to be the primary with the lowest latency/jitter between two points. Your redundant circuit will not be this way. This circuit will NEVER touch any hardware, CO, line across a barren field, that your primary circuit uses. This will mean a slower and lower quality circuit, but you ensure that connectivity will be maintained in the event your primary circuit takes a hit.
Once again even that will begin to affect the cost equation. If management begins to worry about the costs they should be reminded how much it would cost them to have any kind of outage. Todays consumers do not tolerate disruption of services of any kind anymore so the longer your services are unavailable the higher your churn may end up being. There is a big difference between degraded service and no service to the customer. If it's slow they will normally shrug it off unless it's always slow. The second it's down they will begin to agitate.
Since most LANs are Ethernet, WAN connectivity is most efficient by creating an Ethernet internet (lower case "i" meaning a network built out of smaller networks, but not necessarily the Internet (upper case "I") your home PC is connected to ). There is no translation, like with an OC3 and there is very little overhead like there is with ATM solutions.
Most voice applications still use TDM (the OC3 Protocol) and TDM natively breaks any sized circuit into individual 64K voice channels, generally using 56K for payload and 8K for signalling (although ISDN uses 64K for payload and reserves separate channels for signalling). The great thing aboput TDM is that, it is very reliable. If the circuit is passing less than the rated throughput, it is broken and goes into alarm. Under Ethernet, its possible to have a loss of throughput without an alarm.
Both TDM and Ethernet can be oversubscribed by the carrier, however, and equipment can drop information without causing either to go into alarm.
The rub here is that more and more voice applications are using voice-over-internet (lower case "i") -protocol, which is more effective over Ethernet. That is not to say that enterprises are using the Internet (upper case "I", the thing your home computer is hooked up to); it means that they are using the same protocol to deliver voice via private Ethernet circuits.
VoIP is now the main protocol for long distance.
So, if you are a telco with a legacy switching fabric that cannot be replaced, an OC3 makes huge sense.
If you are building a data network or voice network and local carriers have Ethernet available AND the Ethernet service includes your required level of redundancy, then Ethernet is overwhelmingly preferred.
A Cisco fired network can deliver the same 50 ms failover as SONET on 2 fibers with multiple failover routes with very small latency (One network we provision has 1ms in the core and 2 ms each at the serial ports, for a total of five milliseconds of latency). Ethernet can also be designed for point-to-multipoint applications at layer 2 or Layer 3 using very simple equipment. OC3s (TDM) requires use of a large Ethernet router, ATM Switch or Frame Switch with huge operational overheads and the need to program visibility ( the ability for any location to communicate directly with another location).
So . . . An OC3 makes sense if you need more than 45 mbps but less than 163 mbps, Ethernet is not available or is not available in an adequately robust configuration or if your network uses legacy protocols that cannot be replaced.
Price of a complete solution will be based on requirements and specifications, but Ethernet is generally much less expensive.
For more help in finding the best solution for your situation and applications .... whether it's ethernet or OC3 bandwidth .... simply peruse the resources and articles at Broadband Nation.
About the Author:
Michael is the owner of FreedomFire Communications....including
DS3-Bandwidth.com. Michael also authors
Broadband Nation where you're always welcome to drop in and catch up on the latest BroadBand news, tips, insights, and ramblings for the masses.
No. of Times this article has been viewed :
370
Date Published :
Oct 6 2008