Your browser is not supported. Please update it.

13 November 2015

What about a software merger for isolated Juniper?

Juniper must be feeling increasingly like the wallflower at the dance. As its rivals and partners pair up around it, it has failed to secure an Ericsson acquisition (reportedly discussed seriously in the wake of news of the Nokia/ALU deal). It has worked hard to partner with cellular network providers to create some crossover into mobile for its IP business – which has always been more oriented to carriers than Cisco’s, and is highly regarded in that sector.

But its arrangements with Nokia and Ericsson look threatened now by those firms’ new marriages, and Juniper is reported to have been under pressure from activist investors to take itself private. It is certainly running out of potential mergers of its own, with the possible exception of the newly minted Hewlett-Packard Enterprise – unless it takes a radical step and ties up with a software house, reinventing itself as a fully software-driven company and accelerating the process all the network vendors must go through anyway. Oracle, which has acquired smaller IP specialists like Tekelec and Acme Packet in order to reach down closer to the network, might be a good parent for Juniper, not so much for its routers as all the IP and SDN expertise it could bring.

One of the big challenges for network companies in a software-defined era, which Ericsson and Cisco highlighted themselves, is to recreate the OSS/BSS on which carriers’ businesses rest. Without radical change, those platforms will limit the impact of virtualization in other parts of the network, but there is huge risk in disrupting the bedrock of the operators’ business processes – it took most of them many years just to converge prepaid and postpaid operations, for instance.

So why not ease the process, not by building a brand new OSS/BSS, but by joining forces with a player in that area? Ericsson has gone some way down that road with multiple acquisitions in the field, but Juniper could make a bolder move and reclassify itself in the carriers’ minds. Oracle might be interested, while Ron Westfall, research director at Current Analysis, suggests Amdocs, which has capabilities from RAN software (SON, analytics and so on) to OSS, BSS, telco analytics and billing.

For now, Juniper is relying on its own routers and growing software portfolio, and reacted predictably to the Cisco news, saying in a statement: “Our strategy of solving our customers’ toughest networking challenges has not changed. We continue to maintain close business and technical relationships with our partners and customers. Juniper’s solution roadmap is strong and strategically diversified and we are committed to innovating on performance, automation and how we support our customers.”

At its NXTWORK 2015 customer summit last week, it announced the expansion of its NFV (Network Functions Virtualization) portfolio, claiming it could now enable providers to deploy new services more quickly and at massive scale. It has also further disaggregated its Junos network operating system from its own hardware, to support more third party services and applications and enhance programmability. The ability to sell Junos independently (and potentially, for its hardware to run third party software) is a sign of the new software-driven Juniper, and customers will, for the first time, be able to develop and install third party apps directly on to the firm’s switches rather than writing to the Junos SDK (software developers’ kit).

The disaggregated Junos will run on any switches that support the Open Network Install Environment (ONIE), Juniper says. ONIE allows software, including multivendor OSs, to be installed on bare metal Ethernet switches. The software platform uses containers to provide access to the Linux kernel while protecting control, data and management plane functions. This allows guest applications to be run without threatening the network or sensitive apps.

Currently, the new Junos runs only on the Juniper QFX5200 switch but others will be added. The firm’s OCX1100 branded white box switch runs a version of Junos that supports ONIE but is less extensible than its disaggregated stablemate.

On the NFV side, Juniper unveiled Cloud CPE, a line of products to allow cloud operators to deploy virtual CPE in order to accelerate and simplify service delivery and activation or deactivation. Cloud CPE includes the company’s Contrail SDN controller and orchestration tool, and a new appliance called the NFX250. This is an x86 server which lives on-premise and runs multiple Juniper or third party virtual network functions (VNFs).These can be updated or changed without altering the hardware.

To orchestrate them, Juniper announced Contrail Service Orchestration, which manages the VNF lifecycle from design to activation with no disruption to existing applications.

Juniper is relying on NFV to give it a way to outwit Cisco. The latter has committed strongly to virtualization but, of course, has various conflicts of interest with its desire to maintain its traditional hardware base as long as possible. Juniper has some of the same dilemmas but a smaller base to defend, which may give it greater willingness to adopt an aggressive SDN strategy. CEO Rami Rahim said on the firm’s Q2 earnings call in July that Juniper was seeing strong demand for NFV, especially from service providers, and that the technology, along with broader SDN, were “part of pretty much every strategic discussion that we have, particularly with the service provider and sometimes in the enterprise”.

In that quarter, the firm recognized its first revenue on its virtual MX edge router and it is also seeing interest in its virtual SRX firewall, Rahim said. The Contrail orchestration software is key to the effort, and sees Juniper competing with VMware, Cisco and others to provide the dominant networking stack for OpenStack.

But like its rivals, Juniper has to cope with investing in a shift to software while it still gets most of its revenues from its switches and other infrastructure. Rahim, who took the helm in November after his predecessor left under a cloud, insists he is not looking for major acquisitions to turn Juniper around, though he may fill gaps with smaller ones, or with partnerships like the recently announced WiFi deal with Ruckus Wireless. Nor is further major restructuring on the cards, he insists – despite those activist investors.

His strategy is highly targeted innovation, designed to steal a march on Cisco in certain key areas. The overall focus is on agile and scalable cloud-based service delivery architectures, whether physical, virtualized or hybrid, particularly for service providers and government or financial organizations.

There were signs of optimism, after a year of turmoil, when Juniper turned in a better-than-expected first quarter, though analysts at JP Morgan quickly put out a client note warning that product revenue growth would be “modest” in 2015. This is partly because Juniper is rolling out a wide range of new offerings, which will take time to bed in. In Q1, these included the QFX10000 family of spine switches followed by the PTX1000 core router in Q2. The PTX1000 shows Juniper’s difficult balancing act in terms of focusing its development resources and its message – between scaling up its cloud delivery platforms, to be able to handle massive quantities of data, security and applications; and scaling down its physical internet core, an important differentiator since 2012 when Juniper unveiled products to push routers out to the edge, unify wired and wireless access, and even replace some backhaul links. Balancing all that with software is tough, and set to get tougher as its competitors get bigger and fiercer.