Is Juniper Overtaking Cisco in Service Providers?
I don’t know if any of you are getting the same vibe lately but I’m hearing more and more from my colleagues about the service providers they work for transitioning from Cisco to Juniper routers. With reports like Cisco service provider sales slipping 20%, I’ve been thinking about some of the reasons why I see Cisco having a little bit less of a foothold and possibly even losing ground with some service providers. I’ve witnessed this as it has gone down over the years and I think there are a lot of reasons why Juniper is poised and ready in the right place, at the right time. Here’s what it looks like from an engineer’s perspective.
First, let’s acknowledge the greatest agent of change in the service provider space today. Ethernet. Whether it’s upgrading to an all Ethernet core, lighting more multi-tennent buildings with Ethernet switches, back hauling from mobile towers or providing triple play services to a neighborhood, Ethernet is no doubt becoming the distribution method of choice. The increasing needs for more capable and scalable Ethernet devices has a lot of equipment manufacturers in this space cranking out Ethernet boxes as fast as we engineers can evaluate them. One of the most exciting and feature packed Ethernet routers on the market right now is the 1.5 year old Juniper MX. Considering the engineering aspects such as features, software stability, hardware design and ease of use, Juniper really has a leg up on Cisco not to mention just about everyone else with this box. For years, the closest thing Cisco has had is the 7600. Don’t get me wrong, it has served the service provider space for years as a great Ethernet router. The problem is that the 7600 has been plagued with the growing pains of a path blazer with it being one of the first highly scalable service provider Ethernet routing platforms. Let’s take a look at a little history of the 7600 to better understand. From early on, it was pretty much the same as a 6500, which is to say it was an enterprise grade switch that had a router bolted onto it. Between the 6500 and 7600 processing engines were the same, the line cards between them were interchangeable, the same software (IOS) ran on both so there was no real architectural difference between the two. Later, it was decided to fork the two products (which is what they should have done at the release of the 7600) so that there would be different IOS running on each with features for enterprises on the 6500 and features for service providers on the 7600. The 7600 went on to have some unique (and wicked expensive) line cards that added features in the way of VPLS and more granular QoS options that newer Carrier Ethernet services required. There were also two types of IOS created for the 7600, modular and standard. The modular IOS presented the ability to upgrade and restart services run on the system without having to upgrade the entire software package or reboot the router. Creating a modular IOS was done by Cisco to specifically to compete with Juniper’s JUNOS which already had this technology built in. Customer confusion set in as service provider engineers were baffled by what line cards, software and software licenses were needed to provide the services they wanted to give to their customers. With the release of the MX in 2007, Juniper was perfectly placed to pick up where Cisco left off and hit the market when Ethernet services started to explode. Using the same software platform (JUNOS) as they have for the rest of their portfolio of routers makes using it a snap if you’ve already been on their M or T series routers. Even if you don’t have experience and have been an IOS jockey for years, picking up JUNOS is fairly easy utilizing the tools Juniper provides for you on their site such as JUNOS As A Second Language. More on JUNOS later. On the hardware side, the I-Chip 3.0 ASIC used for forwarding on the line cards are truly next generation in that they provide an order of magnitude greater scalability and flexibility for modern Carrier Ethernet services. I guess the point of all of this is, to be positioned as they are with the next generation Ethernet switch/router that has had the past year and a half to become mature is a great position to be in and with them upping their R&D budget despite the economy, I think they are on the right path. The MX will be their flagship product for some time to come especially if they keep out innovating everyone else with real features that service providers really need for their customers. Cisco has answered with the ASR 9000 and to a lesser degree the ASR 1000, but they are young platforms yet to be proven. In addition, believe it or not, Cisco decided to create YET ANOTHER new version of IOS on the ASR 1000. Ugh.
Next, I think JUNOS really has become a powerful sales tool for Juniper with regard to the engineers that actually have to run these things. Unification of their products by using the same software, with the same features and the same commands across the board is incredibly refreshing coming from a Cisco IOS world. Juniper has also had the benefit of Cisco going first here as many of the founding writers of JUNOS came from Cisco where they developed IOS and learned a lot about how not to do things. There are many more powerful command line tools with what seems like a significant amount of thought placed into how people use these routers. There is also a lot more parity to the Unix/Linux command line (you can actually drop into a bash shell if you want to) which a significant number of engineers have experience with. Where I think the real power is in JUNOS right now for the service provider is the XML APIs and OpScripts that are built in to integrate back end operational support sytems (OSS) directly to the router. This is a huge selling point for most service providers trying to automate as much as possible with their OSS as margins get tighter and customer’s budgets shrink in the current economy.
Lastly, service providers are Juniper’s bread and butter. This is the space where they cut their teeth and where they have the most experience meeting customer expectations. I get the impression Juniper has more focus on keeping their service provider clients happy instead of trying to rule everything in every network. Cisco is starting to struggle with their sheer size in the same ways that IBM, HP and Microsoft have in the past. As they add new business units due to organic growth or acquisition, they dilute their ability to perform well in every pie they have their finger in. As an engineer in a smaller regional service provider that has historically been a Cisco shop, I have felt neglected. I have been disappointed by their seeming disinterest in my organization in the way of a string of out of touch sales reps with their own agenda, products that don’t fit my needs and all this at prices that exceed my budgets. It’s for all of these reasons that I’m deploying a Juniper based network to replace my aging Cisco 7200s this year. Yes, I am one of the engineers leading our service provider in what I feel is the great Cisco exodus.
Recent Comments