MPLS in the DataCenter? According to Juniper, they say “HECK YEAH!” and I can totally see why. Most data centers have a nicely redundant, fully meshed switching architecture. Most host servers are multi-homed (or connected to two separate switches for redundancy) and the access layer switches are multi-homed to the core to provide a fast spanning tree for all layer two traffic. And this is great for switching, but move up a layer and more times than not the routing occurs in more spot, at the core on two dedicated routers operating in some sort of failover configuration (HSRP).
Switches have come a long way and even your rack level or IDF closet switches can now route. However, most still switch frames up to the core router to be routed. Most facilities have switches deployed through out the data center that have the ability to route but are just not used.
Wouldn’t it be great if the routing decisions could be made right at the edge? Well Juniper has a clever way to wake up your sleeping routers and begin to add more intelligence between your layer 2 switching and your later 3 routing. Best of all you’re probably using it on your WAN circuits now. It’s MPLS!!
By using MPLS to overlay on top of your OSPF routing protocol within the DataCenter you greatly increase the intelligence and security of your OSPF routing to better optimize your interior data center traffic. I personally don’t fully understand all of the details in how MPLS integrates with the layer 2 protocol such as Spanning-Tree is or how Juniper provides its ciruit to packet resiliency. It could be just vendor fluff, but one thing is for sure, Juniper has been doing provider level services such as mpls and ipv6 for a long time and doing it very well.
I know, I know.. Your saying “Your switches and routers in the data center don’t talk MPLS”. Well with Juniper Network’s EX 4200 Series Ethernet with Virtual Chassis Technology they can! And they’re are listed at a great price point! Between $10-$11K list means you should be able to pick one of these puppy’s up for around $5-6K depending on quantity and your relationship with your VAR.
The EX 4200 series switches offers a full suite of layer2 and layer3 switching capability to satisfy a variety of high-performance deployments including branch and data center locations. This is the same performance that we’ve come to enjoy in other switches like the Cisco 3750 line. But the Juniper EX 4200 goes one step further.
Every switch is PoE. No need to worry about future upgrades for VoIP or wireless AP’s that need PoE power. Every switch enjoys the Juniper Networks Virtual Chassis Technology with allows up to 10 EX 4200 series switches to be interconnected over a 128 gigabit backplane (I believe Cisco can only do 64 gig). Port density will scale from 24 ports on up to 480 10/100/1000 ports. And every switch has the capability to have either Gigabit Ethernet or 10-Gigabit Ethernet uplink options for high-speed connectivity to your aggregation or core layer switches.
All Juniper EX 4200 switches include HA features for redundant, hot swappable internal power and fan trays. Plus the full gamut of JUNOS routing protocols to allow for any standard dynamic routing configuration you can think of. But best of all these little babies will run MPLS to totally hyper charge your data center traffic and bridge your layer 2 and layer 3 protocols for optimum data flow.
I got one of these EX 4200’s to play with and will be writing more later. If these are even close to what has been promised I expect to see phenomenal things.