Wednesday, May 20, 2009

What Bandwidth Solution (T1, DS3, OC3) Makes Sense For A Supply Chain Network?

Firstly, I would probably consider getting a carrier VPN service for the backbone as this removes many of the capacity planning and management issues. You can then concentrate on the bandwidth required to each of the premises and this can vary depending on the circumstances.

For dimensioning the tail circuits, you need to consider the rate of transactions and the typical size per transaction. From this you should be able to work out bandwidth. If you assume largest transaction size and peak number of transactions then this will give you the peak bandwidth. Add at least 20% to this to be safe (TCP overhead will consume some of this). Also consider the service response required. It may be possible to "smooth" some of the peaks by allowing some transactions to be slightly delayed.

Of course you should, ideally, be working this out for each site over a 2 year period and planning for any growth. Find out from your service provider how quickly upgrades can be made and how much they are... it may be better to put in a bigger pipe on day 1. Where it isn't, keep track of the growth rate and factor in the service provider's upgrade time as well as internal delay caused by business case and budget approvals, PO signoffs, etc. and make sure you order in plenty of time.

What you need is a scalable (easily increase Bandwidth for growth), secured (at least as good as layer2 OS equivalent), and flexible (connect various entities that will be part of the SCM). IMHO, I would rather explore what WAN technology would be a better fit, for example MPLS or VPLS than access methods. I truly like the latter because of inherent flexibility, protocol independence, coverage, scalability, security, and yes, flexibility. Some providers may also offer you Secured Internet gateway option thru the xPLS backbone for ubiquitous connectivity and "bring your own internet" connectivity options for smaller B2B links.

DS0, T1, T3 (DS3), OC3, OC12 are all the same technology at different size and price points. The only difference is price ... you buy the size you need. Its like asking what is better to pump water, a half inch copper pipe, a 1 inch copper pipe, a 2 inch copper pipe, or a 4 inch copper pipe ... it kind of depends on how big the pump is and how much water you have.

You don't buy telecommunication for the future, its too flexible - you structure your contracts for the future, you configure your network to expand, you buy the services you need.

I would like to suggest an application level approach to determine your .....

1) application bandwidth requirements,

2) number of locations,

3) extranet/partners requirements,

4) security requirements,

5) closed network or internet based vpns.

Bandwidth doesn't solve all problems, you need a functional strategy to determine your currently capabilities, identify gaps with the solution today, identify gaps and then decide what it would take to address current, and future capabilities, and bridge those gaps.

Companies rely on the technologies in various combination to resolve business problems. they are chosen based on bandwidth requirements, cost, their capability to provide service quality levels, local availability and the degree to which they can interoperate with other technologies and applications.

Given that you supply chain network can have head office, warehouses, suppliers, branch office, wholesalers and retailers accessing it I suggest a multpoint packet switched network as core (sic MPLS).

The IP / MPLS VPN can solve your connectivity problem and you may go for a certain CIR (with time of day, tie of week flavor) that meets your traffic needs to keep cost low. The CIR can range from 1 mbs to Gbps (depending on location) and even remote login can be accomodated. The EIR can be built into the contract to compensate for any traffic fluctuation.

T1, DS3, OC3 are commodity products. It depends on the quality of the network that you need at the locations that you need. Not all services are available everywhere. Also, you might want to consider metro-ethernet, which might be cheaper. With metro-ethernet, you can take into account bursts of needed bandwidth easily. You might want to consider MPLS as well. It also depends on the level of security you need. Often that's dictated by legacy systems, and the customer is not willing to cut-over to a new network because of prohibitive costs or policy.

Although this all may sound very complicated .... it doesn't need to be. You can get no cost help to find the best solution via the services at DS3 Bandwidth Solution

Labels: , , , , ,

1 Comments:

Anonymous Mushroom Networks said...

Hi Michael,

Enjoyed your post. FYI, there is a certain segment of the market that is well served by a new technology, called "Broadband Bonding." This can provide a lower cost alternative, as well as an augmentation, to T1 based transport solutions, including MPLS solutions. It can be used as a solution to bring in additional bandwidth to a facility, as well as increase reliability through augmentation of redundant backup links.

For example, inherently, ADSL is a more modern and efficient technology, from a communication theory point of view. In other words it has a smaller dollar cost per unit bandwidth in some sense. For example, you can often combine 6 ADSL lines and obtain a service that has a greater than 3Mbps uplink and 18Mbps downlink. That's much faster than a bonded T1, but can be delivered for a cost comparable to a T1 line.

For more information, see
http://blog.semo.net/2009/03/31/mushroom-networks-partners-with-semonet-delivering-broadband-bonding-service-to-midwest-business-class-subscribers/

and

http://broadbandfusion.com/

Best wishes.

11:29 PM  

Post a Comment

Links to this post:

Create a Link

<< Home