What to Expect From 'Junisphere'

Juniper Networks Inc. (Nasdaq: JNPR) and Unisphere Networks Inc. are weeks away from finalizing their $740 million merger (see Juniper Nabs Unisphere for $740M). As the dust settles, analysts and other industry observers have been speculating on what will happen next (see Juniper's Kriens: Merging Along).

Two big questions remain unanswered: What will happen to the companies’ competing edge routing product portfolios? And what will happen to Juniper’s relationship with Ericsson AB (Nasdaq: ERICY), now that Siemens AG (NYSE: SI; Frankfurt: SIE) will own almost 10 percent of Juniper?

Neither Juniper nor Unisphere is saying much so far. Both companies declined to comment on speculation regarding the product roadmap.

There’s no question that Juniper bought Unisphere for its IP edge routing. Juniper had focused on the IP core and did a good job in that market. Then it essentially came to the edge market with a core box in a small form factor, creating its M5 and M10 edge routers. As demonstrated by its lag behind Cisco Systems Inc. (Nasdaq: CSCO) and Unisphere in market share for edge routers, this strategy has not been a rousing success.

As a result, Juniper will likely market Unisphere's more successful ERX platform as its premier edge router. But what about Juniper’s own M5 and M10 routers? And what will happen to Unisphere’s new multiservice switch router, the MRX? All three of these products could be in jeopardy, say sources following the company.

"As of Supercomm two weeks ago, no firm decisions had been made," says Frank Dzubeck, president and CEO of Communications Network Architects. "But some products will end up going away."

One potential plan has Juniper repositioning the M5 as an enterprise router and phasing out the M10 router. In this scenario, Juniper might sell the M5 to large enterprises in an attempt to eat away at Cisco’s market share. To do so, Juniper could leverage its newly firmed-up relationship with Siemens. While this idea may sound good on the surface, the logic is full of holes, says Dzubeck.

For one, selling to the enterprise requires a different set of skills and a different sales team. Even if Juniper were to use Siemens's sales channel, it would be a different sales force from the one that Unisphere used to land accounts in Europe. Furthermore, going after the enterprise market goes against everything Scott Kriens, Juniper’s CEO, has preached over the last few months. On numerous occasions he has publicly criticized "competitors," i.e. Cisco, for trying to play both sides of the fence (see Juniper's Kriens Gives Hope).

"The enterprise is an entirely different business and requires a different set of skills and product requirements including software," says Dzubeck. "You can’t just morph a product to target a new market like that. It’s just not the way it works."

Another idea making the rounds is that Juniper is planning to kill Unisphere’s MRX multiservice edge router. The MRX can handle both IP routing and native ATM switching on the same platform (see Unisphere Cutting the Edge). It has a routing capacity of 320 Gbit/s, which is higher than Juniper’s flagship core routers, the M40 and M160. But because of its multiservice functionality, it would likely not be used in the same part of the network as the core routers.

The MRX competes with products from Lucent Technologies Inc. (NYSE: LU) and Alcatel SA (NYSE: ALA; Paris: CGEP:PA), as well as startups like Gotham Networks and Laurel Networks Inc. Some view the addition of the MRX to the Juniper lineup as an asset and a good opportunity for Juniper to expand into another market; but some analysts say keeping the MRX will spread the tightly focused Juniper too thin.

"I’m just not sure that they want to go down that path," says Kevin Mitchell, an industry analyst with Infonetics Research Inc.. "A new market means more competitors and more marketing dollars spent trying to sell it. I just think it would expose them on too many fronts."

Other experts believe that the decision on whether or not to keep the MRX will likely be driven by market conditions. Two or three quarters ago, there was quite a lot of talk among service providers about using multiservice switches to replace private line circuits. Over the past few months, however, that integration path seems to have slipped in priority among RBOCs, says Stephen Kamman, an equities analyst with CIBC World Markets.

"If Juniper sees multiservice switching as a market that is really going to take off, then I’d say they’ll hang onto the MRX," he says. "Otherwise, I can’t see them doing much with it."

Then there is the relationship between Juniper and Ericsson to consider. Siemens is expected to own about 9.5% of Juniper after the merger, and this could cause a conflict between Juniper and Ericsson, its largest partner (see Ericsson: Every Vendor's Best Friend). Ericsson has been instrumental in reselling Juniper routers in Europe and the two companies are also co-developing routing and wireless gear. The first product to come out of this joint venture, GGSN J20 mobile/wireless router, was announced in February 2002 (see Juniper Unveils 'Wireless Router').

Juniper says the reseller agreement will be similar to other agreements it has done in the past. Siemens will have rights to sell both Unisphere and Juniper Networks products worldwide on a non-exclusive basis.

"We don't see any impact on our long term distribution relationship or mobile joint venture with Ericsson," says a Juniper spokesperson. "Ultimately, this is a win win for everyone. Ericsson gets to add the Unisphere product line to its offering, while Siemens gets to enhance its Unisphere edge routing offering with a comprehensive IP solution."

And if history is any judge, this should not be such a big problem. Siemens pulled Unisphere into accounts in many instances, especially in Europe, as in the Deutsche Telekom AG (NYSE: DT) account. But Unisphere was also free to partner with other large telecom companies. In fact, it worked with Ericsson to provide edge routing to Australian carrier TelstraClear (see Unisphere Wins Down Under).

If Ericsson and Siemens can accept each others' partnerships, it would prove a windfall for "Junisphere." The company would then have strong ties to two out of the four largest European telecom equipment companies. Nokia has partnered with Redback Networks Inc. (Nasdaq: RBAK) for edge routing and has an agreement with Cisco for core routing.

"Whatever happens," says Infonetics' Mitchell, "it will definitely be interesting to see."

— Marguerite Reardon, Senior Editor, Light Reading
Page 1 / 2   >   >>
edgecore 12/4/2012 | 10:14:27 PM
re: What to Expect From 'Junisphere'
What are the main differences...aside from more protocols at the edge, more slow speed interfaces at the edge...?


bluemtn 12/4/2012 | 10:14:26 PM
re: What to Expect From 'Junisphere' This is my understanding of the differences, though I may just be stating the obvious.
An edge router needs to be more reliable / stable vs. a core router. When a router has problems in the core it can be routed around, on the edge it causes those customers attached to it to loose their service until the problem is fixed. Because the edge boxes donGÇÖt handle as much traffic, the through put and speed are not as critical as they are for core routers.

The stability / reliability is apparently much harder on the edge because the routers connect to thousands of customer premise routers, any one of which could do some crazy ass thing to cause problems. In the core, your only connecting directly to a handful of other core routersGǪthus on the edge the router is more likely to have problems and those problems are more likely to be known about by the customers.

Design wise this means that edge routers tend trade off some speed for increased reliability.

durtyphiber 12/4/2012 | 10:14:25 PM
re: What to Expect From 'Junisphere' There are a number of differences.

1. An edge box has different reliability requirements. If the edge box goes down, you can't just re-route traffic, because the customer circuits terminate into the box. It's a physical issue. This means that an edge box needs to focus on component redundancy as opposed to be able to use box redundancy, where traffic can be re-routed.

2. Core boxes are focused on switching large amounts of traffic, not on high touch manipulation of that traffic. It's about speeds and feeds. There are some feature requirements, such as congestion management, congestion avoidance, and being diffserv aware. The other focus in the core boxes is fast re-convergence.

3. Edge boxes are focused on aggregating large numbers of customer connections. Where a core box may have a dozen circuits, an edge box may have thousands. This is a different interface mix, and places different stresses on the box. Also, this is where the services are focused. Ingress service level marking (diffserv), traffic shaping and policing, label imposition for MPLS VPN, multicast replication fanout, traffic accounting for metered billing, etc.

kbkirchn 12/4/2012 | 10:14:24 PM
re: What to Expect From 'Junisphere' Core Routers are designed for 100% uptime and optimized for packet forwarding.

Edge Routers are designed for cost-effectiveness and optimized for bandwidth conservation & high-touch features.

Core Router Features:
* Failure Domain Modularity. No single point of failure. All hardware modules can be serviced w/o impacting other modules or the unit as a whole. Main control processor can be upgraded w/o impacting packet forwarding. Redundant main control processor retains full state/route table; no functionality is compromised by a main processor failure or software upgrade.
* Forwarding Performance is consistent & predicatable regardless of packet size, type, route lookup, or hardware configuration.
* High Route Scalability - Router must be able to load full Internet route table and participate as a full Internet routing peer (complete BGP feature support).
* High Bandwidth Scalability - Router must have full range of broadband interface support: (DS3, OC3, OC12, OC48, OC192).
* Core QoS Support - Router must prioritize packet forwarding based on IP QoS classification. Router must not forward packets out of order. Router must signal and intelligently handle path congestion, dropping packets as a last resort.

Edge Router Features:
* Failure Domain Modularity: (Multiple customer edge routers must have core router equivalence, single customer edge routers do not require this modularity.)
* Forwarding Performance may vary depending on features used or packets processed. Consistent performance is not as critical.
* Route Scalability is Limited. BGP is not required. Static routes & smaller-scale RIP & OSPF are used. Most traffic is funnelled through the broadband uplink interface.
* Service Provider Edge: Broadband Interface used as an uplink to the Core router, ranging from DS3 to OC12. Customer-facing interfaces are narrowband.
* Customer Edge: Ethernet interface used as a downlink to the customer's network. Service Provider facing interfaces are narrowband.
* Narrowband Interface Support - Edge routers should support a wide variety of narrowband interfaces including: ISDN, 56/64kbps DDS, T1 Serial/Frame/ATM/IMA, fractional T1, and others.
* Edge QoS Support - Router must conserve the limited bandwidth using traffic shaping, data compression, and application-specific header compression. Packets are classified based on application type and the IP QoS headers modified for use by the Core router. Given the low bandwidth of the links, large packets must be fragmented to allow small voice packets to be interleaved, achieving consistent packet latency for delay sensisitive VoIP.
* Security Features - Router must provide network address translation, stateful packet filtering (firewall), granular access control lists, and VPN tunnel origination/termination.
* Application Proxy Features - Router should act as an application proxy for the most common multiservice network applications (H.323, SIP, Proxy ARP, Directed Broadcasts, etc) to limit the traffic that must be transmitted across the narrowband link.
* Extra Features - Router ideally will support a variety of value-added features such as: DHCP, IP Keyswitch, Protocol Conversion, Protocol Tunnelling, NTP Server, TFTP server, etc....

Iipoed 12/4/2012 | 10:14:23 PM
re: What to Expect From 'Junisphere' Now they need to buy Extreme to round out the puzzle and then maybe the Crisco will finally go back to where it belongs-marketing hype heaven
sigint 12/4/2012 | 10:14:23 PM
re: What to Expect From 'Junisphere' Isn't it the access boxes that would actually link up with customer equipment ?

What could best be described as the boudnary between edge and access ?
whatsup 12/4/2012 | 10:14:23 PM
re: What to Expect From 'Junisphere' looks like they had a major layoff. any
skeptic 12/4/2012 | 10:14:22 PM
re: What to Expect From 'Junisphere' (edge routers)
* Route Scalability is Limited. BGP is not required. Static routes & smaller-scale RIP & OSPF are used.

I think that 2547 VPNs have changed (or will
shortly change) the scalability & BGP
requirements dramatically on the edge.

Marguerite Reardon 12/4/2012 | 10:14:22 PM
re: What to Expect From 'Junisphere' I'm working on a story about that. So please feel free to send the info along to [email protected]
lr_monger 12/4/2012 | 10:14:21 PM
re: What to Expect From 'Junisphere' Edge and Access routers are the same thing, the preferred marketing-speak recently changed from Access to Edge.

Some other differences between Edge and Core are the slew of protocols and functionality needed at the Edge such as:

- IPsec
- Stateful Firewall

These consume lots of processing resources in a router and are hard to scale up to thousands of concurrent connections to users CPE.

And let's not forget:

- Authentications (CHAP, RADIUS, etc)
- Subscriber management (bulk provisioning, etc)

Page 1 / 2   >   >>
Sign In