Hi,

It's still challenging to find tangible information in SDN related news. I just would like to get better analysis of what does this acquisition brings to Juniper.

From all blogs, presentation and  RFC, we can say that Contrail is all about scaling the Data Center using MPLS and BGP as a foundation and addaing some extension. This is a way to claim their approach "open" as it is based on proven standards. They probably have an implementation of thess protocols running in some sort of software. Once can vaguely call this a "controller". It is more likely a BGP Route Relfector with some NorthBound API towards Virtualization Management systems (such as VMWare or OpenStack). This solution needs to run within the DC Network infrastrucutre as well as the DC edge that will connect several DC over operators network.

Juniper is all about JunOS. If you look at what they have been doing for the last 10 years is to expand their OS from SP Core and Edge to new territory such as Entreprise CPE, Switching, Security and DataCenter with more or less success. The Contrail solution is then likely to be ported to JunOS on QFabric Controller within a DC as well as MX fo DC Interconnect. This can be an interesting technical solution. JunOS can provide Open API but will never be Open Source.

But what will be the business impact?

This acquisiiton is related to DC Network Market. I don't see this to generate new business for Juniper. It can simply enhance their Qfabric portfolio. But there's a long way before getting significant market traction in the DC. The major strategic issue is that they are not visible for Entreprise  and DC customers as they are with Operator. Even if they have a techncaly superior solution, they will not be able to get market acceptance unless other major players (such as VMWare, Cisco, IBM, HP...) embrace their solution. It would be really interesting to see IBM and Juniper getting realy close to each other.

The main target for Juniper here could be  the multi-tenant and major operators offering hybrid cloud infrastructure. These type of customer already have a relationship with Juniper and are more receptive to a network focus approach and familiar with MPLS and BGP.

Conclusion:

• This acquisition is good for Juniper to put them on the SDN market craziness map 
• SDN will not change the world, it might be implemented in some part of some networks

Comments are welcome!

I like it. Thanks for chiming in, Kireeti.  :)

Seven is right that the community is a big key to anything open-source. If the community doesn't stay active and engaged, you end up with an 8-track tape player, basically.

But if it works, the community can sprout all manner of innovations that no one player would have thought of.

More pertinient to Juniper, it creates an ecosystem that's not necessarily led by the giants. It sounds to me like part of Juniper's thinking is that Cisco & VMware have an enormous head start simply by already being in the data center. Not something Juniper is going to undo head-on.

You said it!  Thanks, you saved me a bunch of typing :-)

Kireeti.

An observation from Simon Leopold of Raymond James, in his note published Thursday morning:

"...We believe network operators have slowed decisions and requested suppliers to present an SDN roadmap; we suspect this headwind has slowed Juniper’s ramp of QFabric in particular."

I agree with him that it feels like Juniper has been on the defensive regarding SDN. Contrail isn't big, but they've got big plans and are addressing networks of scale (vs. individual data centers).  I do think it'll give Juniper a better SDN story to tell.

I have to say Open Source provides an interesting contrast in the industry.

Smaller Software companies (in particular) use LOTS of open source in products that they sell.  People often ask - So how do Open Source companies make money?  The answer is consulting and service.  Most Open Source software is solid but requires real knowledge to make hum to the maximum.  People sell their expertise (often from inside the organization that placed the Open Source on the market) to help others adopt and adapt the software.

On the other hand most of the major equipment players don't START their projects with understanding what Open Source is available (which is actually how many smaller players work).  There is a lot less control of Open Source, particularly as it comes to security.

I know in our environment we had difficulty with adopting some new versions of software because people put in "fun" patches that were not designed to be in many kinds of production environments and we had to take some parts of these out.  This leads to all kinds of challenges with controlling, testing and updating patches to Open Source projects.  You also have to be careful about the active state of an Open Source community.  Nothing worse than having built a product around what became and abandoned OS project.  Sometimes buggier competitors are better if they remain active.

Which leads us to the last problem.  Remember Open Source does not mean Bug Free.  Open Source projects vary in quality just like any other software.  It generally means that a person building on an Open Source project has to have some form of expertise to debug and fix it.  Those fixes get submitted back to the Open Source team for inclusion in later releases (maybe).  

So, what do you get out of it?  Basically a huge amount of development that you can avoid doing.  Just be clear that your mileage will vary.

seven

The whole idea of open source anything is to simply allow for more innovation at a faster rate than would be possible under the ownership of the market leaders. So in this case, will it even matter at all that Juniper takes an open source approach to SDN?