Why did Juniper buy WANDL?

Is it part of an SDN strategy? A knee-jerk reaction to Cisco buying Cariden? OSS Line collects a number of industry commentators’ points-of-view on the acquisition of WANDL by Juniper.

When a company like Juniper buys a small software vendor it creates interest and speculation outside of the OSS industry. And when a ‘hardware’ company buys a network planning ‘software’ company, there must be an SDN story there, right?

The acquisition for $60 million follows two other purchase of network planning solutions just over a year ago: Cisco bought Cariden for $141 million and Riverbed bought Opnet (who also offer a lot of application performance management tools) for $1 billion.

I have my own opinions, which I’ll leave to the end of this post. First, let’s see what a few other industry experts think about this…

Most of the main-stream OSS and telco press did little more than re-post the official press release, butRay Le Maistre of Light Reading  did have this to say:

And as SDN creeps into wide area networks, the ability to manage a mix of traditional, hardware-based network elements and software-defined elements is going to be key in the coming years, so the combination of its Contrail SDN solution with enhanced management and optimization capabilities from Wandl looks like a sensible move by Juniper.

By the way, it’s WANDL not Wandl. It’s an acronym for Wide Area Network Design Laboratory.

Tim Nolle, on the CIMI Corp blog, gives us two detailed posts about his thoughts on WANDL and Juniper.

In Does Juniper Have a Magic Wandl? he considers a number of possible Juniper strategies that might be behind the deal. Is Juniper starting to build a next-gen OSS software stack? Unlikely. Are they buying WANDL to add an SDN/NFV-capable analysis/control layer to their NMS offering? I’d say, quite possibly. Are they bringing a technically valuable partner in-house to mitigate the risk of Cisco or ALU snapping them up too? Almost certainly.

Tim followed this up after some debate on LinkedIn with his Wandl and Cariden: Is There a Real Value? blog post, asking:

…whether policy management and path computation might be a factor in the deal, even to the point where they might be incorporated into Juniper’s Contrail SDN controller as a result of the buy.

So, a logical step beyond using WANDL to analyse and configure Juniper’s SDN networks would be to use WANDL’s PCE (path computation engine) as a cornerstone of the SDN controller. The challenge here would be taking an off-line optimisation process and deploying it as a real-time on-line network control process.

PCE has been around for a while, and has seen only modest adoption by IP/MPLS network operators. Does WANDL offer something that might increase its value in an SDN network?

…if we presume that SDN devices, lacking adaptive behavior, might also lack some of the traffic knowledge that’s necessary to do adaptive routing, you could argue that SDN might need [WANDL’s] traffic knowledge more than legacy networks would.

Interesting point. If SDN devices are simpler (because routing becomes centralized in the SDN controller)  than current IP devices (which are pretty simple but at least need to know a bit about the neighbouring devices and end points to determine the next hop for a packet) then the SDN controller needs to be smarter, more network-aware. Which is exactly the role of WANDL and other similar solutions (Cariden, Opnet, Aria Networks) in legacy IP/MPLS networks.

Back to the press andPaula Bernier at TMC’s SDN Zone writes:

[Juniper Networks’] most recent acquisition ­– of WANDL for $60 million – is already paying returns. According to reports, Juniper’s stock is on the rise following the deal, which positions the company to offer more affordable and flexible networking solutions.

The traders must know more about Juniper’s strategy than us experts. Bit embarrassing…

Paula quotesPeter ffoulkes of TheInfoPro, whose generic analyst statement about Juniper’s future in SDN/NFV/Cloud luckily hits the spot regarding WANDL:

As organizations move beyond virtualization of production workloads, attention is shifting toward the management and automation of the software-defined data center.

What do I think? While I have not worked with Juniper,  I did spend the last four years in a business competitive with WANDL and informally talked about the whole planning, optimization and SDN thing with the guys working for various IP hardware vendors. So I don’t have the inside track on Juniper, but I do have an informed opinion.

As almost everyone points out, Juniper and WANDL had an existing partnership. But this was not exclusive. Juniper also worked with Cariden and, as I understand it, they were probably the favoured partner in IP/MPLS planning. Cariden certainly has a stronger product in this area, with WANDL software coming from the days of ATM and being re-purposed for IP/MPLS. WANDL also worked with other hardware vendors. They were, until recently, listed as Alcatel-Lucent Connected Partners (along with Opnet, Cariden and Aria).

Up until this year the hardware vendors were happy with this little, competitive and independent IP/MPLS planning OSS ecosystem. The hardware guys were happy selling devices that could do all sorts of clever things with IP and MPLS, doing multi-million dollar deals and leaving the complexities of managing the networks to the software guys and their couple-of-$100k-deals. Everyone was happy.

Then Cariden and Opnet got bought.

Who was Juniper going to bring in to offer MPLS network planning and management if their other partner, WANDL, also disappears in to a competitor?

And this happens at a time when Juniper is building an SDN strategy that, as Tim Nolle notes, could well be delivered by leveraging their existing IP/MPLS capabilities, supported by some smart analysis, planning and optimization processes.

The risk of being left with no IP/MPLS software partner was too great. And the cost of acquiring WANDL was so modest. Secure the partner, and Juniper secures a significant chunk of their future software strategy.

This isn’t Juniper moving in to next gen OSS, but simply the necessary software to continue to supply IP/MPLS now and SDN in the near future when there is no longer a reliably independent ecosystem of software vendors.

, ,