56K to T1 upgrade Frame relay

[chang-michael]

I have never gone through such a upgrade. I have over 20 remote sites and we would like to upgrade all the sites to t1/ frac t1. We're also upgrading the routers at the remote sites from 2500 to 2600s. We need to do these with minimum downtime. I was just wondering if anybody had experiences with frame relay providers how such a cutover can be done. The issues are that do we get a new PVC for each site for the new t1/frac t1 circuit. Appreciate any suggestion / experiences on this.

One other thing.. Redundancy has come up has as issues and we're looking for a 2nd frame relay provider as a backup. How easy / difficult is to get a fully diverse T1 from a site and ensure that it takes a different route, CO, etc. Has anyone been successful doing this ?? Many thanks

[steven-dunn]

If you have the right NTU's (some NTU's can only handle up to 128K, some to 384K) the upgrade should take around 10 minutes per PVC. The service provider should be able to do it all through software changes. If the carriage hardware can not support these speeds you will need a longer outage to replace the SP hardware. You probably won't get new PVC's or DLCI numbers.

Unless you install a completely new frame access, cutting over to the 2600's will probably take 15 minutes each site, mainly just for verification. Do it in the lab first. If you are installing completely new accesses, you can run the 2500's and 2600's side by side and cut them over after the carriage has been verified, outage of about 1 minute.

You need a realy good relationship with your SP (lots of $$$) to achieve diverse feeds. At the end of the day, both frame services will enter your building via the same place anyway so complete diversity is rarely achived. Ask your SP for SLA's and ensure they abide to them.

[chang-michael]

Thanks for the respone. Today there are 56k DDS circuits, so obviously there will be a new T1 CIRCUIT. We're also putting a new router, I am how the SPs actually manage the transition. Do they provide new DLCI/PVC and delete the old one once the cutover is done. Regarding diversity, I know that fiber providers will provider fiber to the premises through diverse paths / entry into the building, so that is fully diverse. Dont how T1s are done ?

[steven-dunn]

My SP can deliver both DDS and Frame on the same NTU as long as it is not completely subscribed. If you are going to use the full T1, they will install a new T1, allocate new DLCIs which allows you to build the new network and completely test it before cut-over. They will then deactivate the dds later on.

If you can get your providers to supply diverse fibre, they can then connect a MUX to the fibre in your building to deliver your T1 tail . generally, the MUX is in the exchange and they deliver the tail to your premisise via copper.

[thisisshanky]

Hi,

Since your previous network was on a 56k, i think you willhave to order new pvcs. the provider will provision the new pvcs and install the new smart jack at your remote sites. Once the cut over is done, you can disconnect the 56k service.

Are you planning to use external CSU /DSU or the use wics with internal csu dsu on the router ?

You can preconfigure the new 2600 routers with the new dlcis and other equivalent configurations and schedule a downtime to install the new router and perform the cutover.

For ordering FR circuits with most of the service providers, can be done over the internet, if you signup for a corporate account with them. Best approach would be to contact your service provider provisioning agent.

To have a 2nd frame-relay provider is a good idea, but then it might prove to be costly, because two independent circuits wil have to be ordered to achieve this.

Another way to achieve redundancy would be to create two pvcs from each remote site to core network. One pvc can be primary with a higher CIR, with the other pvc being the secondary with a lower CIR. Now both pvcs can terminate at a core router at the HQ. (if you want router redundancy at the core you can make these two pvcs terminate at two core routers)

A routing protocol such as EIGRP would help in faster convergence in such a network.

The above suggestion doesnt provide a service provider level redundancy, though.

You can inject static default routes to the remotes and keep the whole routing table at the core routers.

If you need more help you can personally email me at snair28@yahoo.com

HTH