SD-Wan Implementation

[jemimajapay9]

Hi, Good day, 

I am just new in networking side.

I would like to ask about the possibility of implementing SD-WAN in our company.

We currently have two branches connected to a main server. The main server is located at our main branch, which has its own ISP connection, and we are using a point-to-point (P2P) link to connect the other branch. Recently, the other branch also acquired its own ISP, to ensure connectivity in case the main branch goes down.

My main question is:

Is it possible to implement SD-WAN so that both the main and branch office connections can be unified to improve overall speed, stability, and failover?

We are looking for a way to make the connections work together, not just for redundancy but also for performance enhancement. Is this achievable through SD-WAN?

[Joseph W. Doherty]

Unfortunately, I don't know the correct answers, although I suspect SD-WAN might be able to.  If it can, I suggest your next question should be, should it be used?  The answer to that will likely also depend on what exactly you have now and your network desires, along with, possibility, the need to upgrade your network to fully leverage the benefits SD-WAN might provide.  I.e. would its benefits (if any) outweigh its additional costs (if any).

If the answer is negative, do keep in mind the technology continues to evolve, so a negative, now, might be a positive within a few years.

[MHM Cisco World]

SD-WAN is perfect for hub and spoke even if hub have two ISP.

MHM

[Joseph W. Doherty]

SD-WAN is perfect for hub and spoke even if hub have two ISP.

Laugh, it may well be, but the ancient caveat emptor, often is still good advice,  Also, the much more recent, "I'm from Missouri" has some value too.

[MHM Cisco World]

For any reason' dont mention me again.

I Dont have time to waste

MHM

[Joseph W. Doherty]

Ah, if you want to avoid wasting your time, don't post then, although many of your posts contribute much, so, personally, I don't think you're wasting your time.

If your actual request is you don't want me, personally, to, in anyway, reference anything you post, sorry, unlikely to happen, as long as you continue to post and I follow the "rules".  I don't believe contrary opinions or requests for additional information are prohibited.

In this specific case, and all hub/spoke topologies, you assert SD-WAN would be perfect.  I agree it may be, but I'm of the opinion such a statement often benefits for some explanation why that's so.

Possibility what you found upsetting is my "laugh".  Well, possibility I shouldn't have written that, but over the decades, the many times I've heard some variant of something being perfect, or that's nothing to worry about (usually by sellers), usually has been found not to be the case.

So, this kind of laugh is the "grin and bear it" kind, not it's a funny joke kind.

A good chunk of my career has been involved in "fixing" issues so often caused by rushing a "solution" without due consideration of the "problem".

So, to summarize, in this case, you believe SD-WAN is "perfect".

Again, it may be, but I'm of the opinion, perhaps, some additional analysis of pros and cons should be undertaken before jumping into SD-WAN.  Further, even if such an analysis is negative, in a few years, it may become a positive.

[jemimajapay9]

Hello Sir,

Do you have any recommendations for our network that could help us achieve faster internet? We’re experiencing issues with video conferencing—it feels like the connection is slow. Our second branch is 2 miles away from the main office, which is why we’re using a point-to-point (P2P) link to connect both sides. This setup also allows the second branch to access local resources from the main office.

[Joseph W. Doherty]

Recommendations?  Possibly, but "Internet" and vidconf usually, and fundamentally, don't play well together (BTW, I don't see how SD-WAN can guarantee a solution.)

The fundamental problem is vidconf, to work well, needs specific network performance.

On something like your p2p link, that can usually be easily guaranteed with QoS, which can be applied on both ends.

On an Internet connection, we can apply QoS on our side, but not to the far side we don't control.

(Technically, there are some our side techniques that can influence far side transmission rates, but typical routers don't support them, and they are far from being foolproof.)

That said, is vidconf and Internet incompatible?  Not necessarily.  In your case, would need to know specifics of your topology, and exactly your network usage needs.

Also, for some requirement/desires, like "faster Internet", a typical answer might be obtain more bandwidth, which might actually be needed, but there can be other causes and/or effective changes too.  For example with Internet bandwidth, inexpensive gig isn't uncommon, but can your Internet devices handle it, and Internet bandwidth might not be full port speed, and ISP drops bursts which you might have buffered.

[MHM Cisco World]

Sdwan use AAR for voice/video traffic' 

So use sdwan I think is good for your requirements

https://www.cisco.com/c/en/us/td/docs/routers/sdwan/configuration/policies/ios-xe-17/policies-book-xe/application-aware-routing.html

MHM

[Joseph W. Doherty]

As an analogy to why I suggest additional analysis of pros and cons of selecting any new technology, for example, perhaps SD-WAN, consider:

Choice of how routing is done, i.e. static vs. dynamic, and with the latter, all the various dynamic routing protocols.  Further within a dynamic routing protocol, like OSPF, do you something like its area capabilities?

So far, we know, you have just two sites, the primary with servers.  The sites are interconnected by a p2p and both sites have ISP connections.

Given only that I information, personally I would think it a tad premature to suggest something like MPLS, BGP, segment routing, etc. is perfect, it's just what you should migrate to.

Now assuming any "advanced" technology would truly "improve" your network, and it may, is the improvement worth the additional cost?

Or, say, I suggest every host should have a  dedicated 100g p2p link to every other host.  Many might consider such a recommendation ridiculous, but technically, you cannot get much better, well perhaps other than having multiple 100g links between hosts.

The last is an extreme, but perhaps the Pentagon, some huge investing/banking firm, some major research project, etc. can validate such a network, for some specific case.  Likely, though, they have done some analysis, and considered pros and cons, for different alternatives.

Or, mundane, I don't suggest a hammer or screwdriver without knowing a little more about the situation.

Again, SD-WAN might, indeed, be perfect for you, or in might be possibly an expensive migration that provides no noticable real world benefit.  IMO, to date, insufficient information to say.  Likewise, cannot suggest the value of alternatives.

I believe I've mentioned I've not used SD-WAN, but I was an early implementor of Cisco's precursor technology, OER and PfRv1.  I also have about a decade figuring out how to use QoS, effectively for all traffic.

Basically, for about a decade, my primary role was making an International network actually work well, without buying international WAN bandwidth in 100g lots.

Also BTW, as much as I believe QoS is underappreciated, there are cases where additional bandwidth is necessary or it's the more cost effective solution.

So, I'm not at all against using SD-WAN when it's the best solution, all things considered, but, IMO, we're far from having considered all things in your case.