I'm having a discussion with a small network in a part of the world
where bandwidth is scarce and multiple DSL lines are often used for
upstream links. The topic is policy-based routing, which is being
described as "load balancing" where end-user traffic is assigned to a
line according to source address.
In my opinion the main problems with this are:
- It's brittle, when a line fails, traffic doesn't re-route
- None of the usual debugging tools work properly
- Adding a new user is complicated because it has to be done in (at
least) two places
But I'm having a distinct lack of success locating rants and diatribes
or even well-reasoned articles supporting this opinion.
I'm having a discussion with a small network in a part of the world
where bandwidth is scarce and multiple DSL lines are often used for
upstream links. The topic is policy-based routing, which is being
described as "load balancing" where end-user traffic is assigned to a
line according to source address.
In my opinion the main problems with this are:
- It's brittle, when a line fails, traffic doesn't re-route
- None of the usual debugging tools work properly
- Adding a new user is complicated because it has to be done in (at
least) two places
But I'm having a distinct lack of success locating rants and diatribes
or even well-reasoned articles supporting this opinion.
Am I out to lunch?
No, but what better solution do we have to offer them? There are dynamic
load distribution features and products (think Cisco PfR, for example), but
those are routinely lambasted as well.
* Is a fragile configuration. You're typically forcing next-hop without
a [direct] failover option,
* Often incurs a penalty (hardware cycles, conflicting feature sets, or
outright punting to software),
* Doesn't naturally load-balance (you pick the source ranges you route
where)
However, there are few alternatives in some cases...
* If you are using some provider-owned IP space you often must route to
that provider,
* There may be policies restricting what traffic (sources) can transit a
given provider
There are few alternatives for the latter cases, unless you split the
border across VRFs and assign routing policy on the VRF, which is a
global decision across the VRF, and avoids PBR.
We're doing a little of both, so I clearly don't take sides
