Bonded T-1's to Multi-T-1's - what impact on network traffic, configuration, etc.

We're a group of medical clinics that need to move large image files (3-D PET, MRI, CAT etc.) between clinics.  

We can get Bonded T-1's in our metro area, but some clinics are in rural parts of the state (and even 2 in other states) and do not have Bonded T options.

What can you tell me about performance, trade-off's, configuration and management on this type of network?  Will it be worth it to have multiple T's at the remote clinics if that cannot be 'bonded'?

We run VPN's over the Internet using cisco equipment - so all T's are from various ISP's (same in metro area, different in rural area) - we have no private ppp t-1's.

I'm sure I'll have more detailed questions on configurations later - but will post those separately.

This is primarily an information gathering exercise

Who is Participating?
I wear a lot of hats...

"The solutions and answers provided on Experts Exchange have been extremely helpful to me over the last few years. I wear a lot of hats - Developer, Database Administrator, Help Desk, etc., so I know a lot of things but not a lot about one thing. Experts Exchange gives me answers from people who do know a lot about one thing, in a easy to use platform." -Todd S.

My two cents:

We have two T1s for our WAN and two T1s for Internet.  You'll still obviously get a performance increase just by having the additional lines.  Someone else can speak better on the bonded multiplexed lines, but our two Internet T1s come into the same router and we do a packet load balance which works great, achieving the 3MB.  For the WAN, we use two separate Cisco routers and use GLBP to load balance those two T1s, still offering us the 3MB.  That also offers some redundancy so we can lose a router and the line stays up.  We just lose the extra bandwidth.  It doesn't do quite as good a job as per-packet load balancing (by design), but it still works just fine and reduces our potential downtime.  
The main reason for bonded connections is to get more speed.  The "bonding" makes the datalink speed the equivalent to the sum of the speeds of the bonded links.

The main reason for multiple lines is redundancy/reliability.  Usually you get service from more than one provider and so if one goes down you are still (possibly) up.  You can load balance on multiple lines but the maximum speed will be the speed of the individual line that the traffic is flowing on.

What is best for you depends on what problem or situation you are trying to resolve.
If they are not bonded they can only be used to enhance your environment if you assign them roles and use clever settings on your routers and your workstations'/servers' default gateways.  For example you could have one T-1 for all mail in and out, and the other for web browsing.  Manually splitting the duties.
Ultimate Tool Kit for Technology Solution Provider

Broken down into practical pointers and step-by-step instructions, the IT Service Excellence Tool Kit delivers expert advice for technology solution providers. Get your free copy now.

dsbahrAuthor Commented:
If I have 2 bonded T's I get a single 3 meg pipe, but 2 multi-link T's I get 2 1.5meg pipes, but load-balanced with possible redundance/failover -- is this a correct summary?

Also - there should not be any problems with bonded T's at the corporate center feeding multilink T's at a clinic - it will just throtle down to the max speed of the slowest link - correct?

For a large file transfer (say 40 to 400 Meg) will I get any improvement in transfer time with a multi-link T-1?, or is it simply that the file xfer will take one T, and all other traffic (terminal services, e-mail, telnet, etc.) takes the other?  Would this happen automatically - i.e. the router knows one is full and shunts everything to the other until finished - or is the router 'dumb' and it tries to feed new data-streams round robin with now knowledge of existing load???

We used to have four 512Kb bonded ADSL lines linked in a Cisco router.  This provided a much faster link for browsing and file sharing, etc, even exchange traffic.  

However, for somthing that is mission critical and requires a guaranteed stable connection I would not recommend it.  We had serious problems with Terminal Services clients connecting via the bonded lines, meaning we couldn't function as a business.  The bonded lines were scrapped for a more expensive but more stable leased line.  
Your summary is correct.

You should see improvement with a multi-link T1 since it will consume additional bandwidth now.  However, the file copy would most likely not use just one T1, and the other traffic would use the other. If you setup packet load balancing, it goes round-robin and will evenly balance the total load.  Using GLBP, on the other hand, is circuit based.  You get less-even balancing, but a connection tends to stay on the single T1.  As Carl Legere eluded to, you can specifically control the traffic and assign it to a T1, but I never favored that.  I just balance the total load to take advantage of my total bandwidth.

Perhaps to further clarify, I don't need bonded T1s to consume 3MB bandwidth.  If I kicked off a huge copy or transfer, and I have two load balanced T1s, I could potentially saturate both T1s and use my full bandwidth.
I don't favor it either, but there isn't much you can do without BGP or something on the ISP side.

usually wherever you can get a t you can get a provider that supplies hardware that can multiplex them together and hand off ethernet to you.  Packet switched (T-1) is technically avalible everywhere for a price.

I am not infavor of bonding T-1's since there is new CO hardware that does it for you.  Many providers are switching to this hardware from overture.  

Experts Exchange Solution brought to you by

Your issues matter to us.

Facing a tech roadblock? Get the help and guidance you need from experienced professionals who care. Ask your question anytime, anywhere, with no hassle.

Start your 7-day free trial
Time out.  The OP is doing a VPN.  All packets will likely have the same source address (local firewall) and destination address (remote firewall), potentially on rarely-changing ports (GRE, UDP/500, TCP/10000, etc.).  This will be VERY hard to load-balance over multiple NON-BONDED T1s at a given site.  CEF per-packet load-sharing can cause out-of-order packet delivery; some applications (VoIP, perhaps VPN) can really get cranky when OOO happens.

Now, can you clarify: you say some of the sites can't get bonded T1s.  Does that mean you'll only get one T1 to a location?  If so, you're fine, as long as the bandwidth is OK.
dsbahrAuthor Commented:
Thanks for the info everyone!

I've been need-deep in aligators the last few days, but appreciate the info.

I will next post for actual configs with cisco route and pix's across a mix of bonded T-1's and different carriers.

It's more than this solution.Get answers and train to solve all your tech problems - anytime, anywhere.Try it for free Edge Out The Competitionfor your dream job with proven skills and certifications.Get started today Stand Outas the employee with proven skills.Start learning today for free Move Your Career Forwardwith certification training in the latest technologies.Start your trial today

From novice to tech pro — start learning today.

Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.