• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

IMA vs. Load-Balance

P

polm

Diamond Member
Lets say, for example, that I need multiple T1's (3mb - 10mb) at a location.

What would be the overall difference between using IMA to multiplex the traffic , or just using a routing protocol that supports equal path load balancing?

Is the advantage the ATM functionality of IMA ? (PCR/SCR for example)
 
In this case I have 1 datacenter in Texas, and 1 datacenter in Atlanta. These 2 Datacenters provide redundant services. Remote sites will forward all external traffic to one of these 2 datacenters.

Let's take one of those remote sites; 3mb of bandwidth is required to support the remote site. 2x T1's will be ordered.

With IMA, we would create 2 seperate PVC's toward each datacenter and still configure load balancing across the 2 PVC's.

Without IMA we would bind 1 Frame Relay PVC to each T1 toward each datacenter and configure load balancing across the 2 PVC's.

Considering our network is built upon a ATM/Frame cloud and all remote sites require redundant head-ends at our 2 datacenters, would multilink PPP still be an option?



 
well you didn't mention you were doing atm/frame.

In that case IMA may be your better option.

The great thing about mlppp is it appears as one single layer 2 connection (and IMA as well.) Both technologies will give you true "combined bandwidth". I'm not a big fan of CEF for load balancing as it never seems to provide the "add the pipes together" performance of IMA or mlppp.

But your mention of PVCs concerns me. It tells me these really aren't point 2 point T1s, but are frame-relay circuits delivered on a T1.

if you have data centers backing each other up that application screams dual T3s and MLPPP.
 
Originally posted by: spidey07

But your mention of PVCs concerns me. It tells me these really aren't point 2 point T1s, but are frame-relay circuits delivered on a T1.
Click to expand...

Well, in some cases the site will have a single T1 line purchased with 2 PVC's.

It all depends on the capacity requirements. Some sites have 2 seperate physical T1 circuits with only a single PVC assigned to each seperate serial interface.

Once we move into 3+ T1 requirements, then we always go IMA with 2 PVC's.

My major question was in the case of our remote sites that require 2x T1's only.

if you have data centers backing each other up that application screams dual T3s and MLPPP.
Click to expand...

I'm not sure what you are saying...😕



 
But your mention of PVCs concerns me. It tells me these really aren't point 2 point T1s, but are frame-relay circuits delivered on a T1.
Click to expand...

Interesting point. I guess the PVC over T1 isn't the same as a dedicated T1.

In the case of sites where we have a dedicated T1 line, we usually create 2 GRE Tunnels back to each datacenter.

I have not installed a site with multiple dedicated T1 lines (no PVC), yet.

 
I'm saying that redundant data centers should be operated with Dual T3s, point-2-point with MLPPP.

You're in the "tough spot" with needing greater than T1 bandwidth. what technology you can use if you have a frame relay network is totally dependant on your provider. I'd talk to them. Not the sales guy, request to speak with a design engineer.

You're request is very common and can only really be answered by the provider.
 
Not only is this "tough spot" difficult to manage, but my company has been "merger crazy" in the past few years and we still have a lot of different implementations going on. From point-to-point leased T1's, to Frame PVC's, and IMA...and this is just the small remote sites.

Call centers and large business offices are a completely different story...always ATM over DS3, using vbr-nrt, and redundant PVC's back to the datacenters.

But when it comes to these smaller sites, and I make a request for bandwidth installs/upgrades I never know exactly what to expect. I'll get the bandwidth, but it comes in a few different flavors.

I'm just trying to understand all the different technologies. I haven't seen any MLPPP, yet.
 
You may also be able to get Multi-Link Frame-Relay.

As before, check with your provider to see if it's available.

MLFR is usually less expensive than IMA, doesn't require any serious hardware upgrades (one port per T1, probably IOS upgrade ... depending on your router platform, you may want to upgrade that ... should be 2600 or better, 28XX would be an excellent choice).

IMA generally requires higher-end hardware and tends to be much more expensive. Because of the overhead, IMA with only two T1s is not really buying you that much extra bandwidth for the money.

MLFR has a bit less overhead (you're still dealing with frames versus cells), still uses DLCIs, is fairly easy to expand for future growth (again, depends on your base router platform), and is / can be a bit easier to troubleshoot (at least on the consumer end of the connection).

As a general concept, MLFR is roughly comparable to MLPPP, but the base transport is Frame Relay instead of HDLC.

Check it out. It's fairly new, compared to IMA offerings, but most providers using Cisco, Lucent, or Alcatel switching infrastructure should be able to support it

Good Luck

Scott
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top