Monday, October 06, 2008

Is OC3 Bandwidth Still A Viable Option For A Network Infrastructure?

It appears that ethernet based configurations are fast becoming the option of choice for larger network infrastructure applications .... over OCx based solutions. However .... when, where, and why would OC3 bandwidth solutions still be viable and/or prefered?

An exact answer to "when, where, and why" would depend on the demands that would be placed on the network and the availablility of existing resources. If there are already SONET OC pipes in place then it would make sense to utilize them for a network connection instead of installing new facilities. What would be a major driver is the bandwidth utilization that might be placed on on the network connection. With the trend for more bandwidth intensive applications an OC3 pipe might be quickly used up once your network grows beyond a small segment.

Generally speaking OC3 would be preferred over ethernet when/where/why:

- distance greater than 300ft (up to 3km?)
- high electrical interference environment
- slightly faster
- lower maintenance

Generally why you would prefer ethernet over OC3:

- easier to manage/support
- more options for inter-connectivity devices
- less expensive
- much much less time to implement

SONET (e.g. OCx) networks have the following advantages over all packet networks.....

1) Protection: Restoration of services after the detection of a fault is done within 60 mS. All packet networks can do this as well, if an RPR architecture is implemented. Restoration capability in SONET networks comes at a cost; 50% of your bandwidth is sitting idle. RPR reuses the spare bandwidth resulting in 100% network utilization.

2) Network Management: SONET systems have much evolved and hence powerful feature sets in their vendor provided NMS's.

3) Performance Monitoring is also more evolved in SONET systems and facilitates troubleshooting greatly.

4) Synchronization: Back haul applications like cell tower requiring accurate synchronization currently have less issues deriving clock from SONET networks than they do from Ethernet based networks.
Timing solutions like IEEE 1588 exist for Ethernet networks and evolving to address outstanding issues.

Despite the fact that OC3 equipment is very cheap, I would implement any new infrastructure using all packet technology. The scalability can be done on the fly in fine (1 Mb/s) and coarse increments (10, 100 Mb/s), unlike SONET networks where you are stuck with T1, T3, OC3 and OC12 increments, etc. Ethernet technology is ubiquitous and is way of the future.

If I had the choice, I would just go with ethernet...especially with 10Gb well on it's way.

These days there is so much overlap of purpose because of the force fitting of Ethernet concepts into OC frames. Some of this bastardization makes sense, such as VCATing to more efficiently utilization of bandwidth when Ethernet is the source or destination for the signal. But, honestly, most of the bastardization is not for "good reason", it's for religious reason.

It's because most network engineers are not really "engineers" but rather, "network" guys that took an A+ test and don't know care so much about what's best, but what they know. Hence, religion. It's the same as Java vs. C# or VHS vs. Betamax. It's not so clear whether one is better than another, but rather which people are more comfortable with.

Back to the real question. Yes, OC is a viable option, but since 10Gb Ethernet is (more or less) interchangeable with OC-192, and most engineers know ethernet better than OC, it doesn't make a whole lot of sense for most networks to build out for OC. However, if the network will be buying time on existing SONET infrastructure, then picking up equipment that is at least compatible, even if more expensive, is worth while.

Simply put, if your service provider has that infrastrucutre in place (OC3) that is what they will push. If your service provider has optical ethernet in place, they will offer this technology. Technically optical ethernet is better as a delivery system (voice, data, video) but "Bell Heads" would argue that point. The OC3 hertitage is one to support channelized voice streams, optical ethernet for data streams. But with the convergence of voice and data in transport methods for supporting business applications .... ethernet has an advantage.

By the way both are optical methods so the "glass stays the same". Only the equipment you attach to the ends will differ.

OC3s, and Sonet in general, are all still very viable (and have several advantages). However I am a fan of Metro Ethernet solutions .... or even national point to point circuits for delivery of bandwidth, where available. Metro-E is generally cheaper, often both for the provider as well as the customer, and thus it is quickly becoming available in many markets.

Sonet applications have one advantage over some common Metro-E offerings that you should be aware of. With Sonet you are guaranteed that your interfaces will pass link state end to end. This could be important if you are running OSPF or some other IGP that could take advantage of the extra information to more quickly route around failures. (and yes, there are technologies that can detect link failures even on media types that do no support passing link state).

There is no doubt that in major metros, the flexibility and fact that ethernet based technologies can be managed by traditional network staff and do not reguire a specialized telco engineer is certainly a benefit. Costs seem to favor ethernet over OCx offerrings as well.

Doesn't it depend on the application running across the service though? If I am doing asynchronous replication of a WAN to a secondary location, I still may need OCx technology. I may need other options as well. FC over ethernet is out and it does open possibilities, but the price points are still very high and is this a tried and true technology yet?

In the end, many areas are constrained by service offerrings. In areas where there are options, it still comes down to the needs of the customer based on the application(s), the experience of the staff managing the solution, and price points.

While Ethernet is a great alternative for many of today's networks, there is still a strong marketplace for OC-3 services, particularly when you look at the price of an OC-3 compared to 2, 5 or even 10 years ago. Also, unless there is a dramatic requirement for migrating to all Ethernet based services, existing OC-3s retain money already spent on infrastructure hardware and it is still relatively easy to troubleshoot. For newer networks, let your applications and services drive your requirements for Ethernet based services, at which point it comes down to cost versus functionality.....

For help in determining the right solution .... whether it's ethernet, OC3, or something else .... I strongly suggest taking advantage of the no cost support available here: Bandwidth Solution

Labels: , , , , , , ,


Post a Comment

Links to this post:

Create a Link

<< Home