Cisco & Juniper - Two most competitive companies but from last few years Juniper has been knocking out Cisco in every Arena
(being a Cisco fan boy, I cant explain how bad I feel about this). Cisco & Juniper are fighting hard to prove their competitiveness in Datacenter Switching and both of them had released revolutionary products like
Nexus &
QFabric. We had talked about both these products earlier and both these products stands neck to neck (Okie.. QFabric a bit superior one). Both these products support Fabric Extenders -
Fabric Extenders are individual swicthes generally used as Top of the Rack (ToR) switches but when connected with main chassis, these swicthes can act as an extended line card.
So, Lets spend sometime today talking about these
Fabric Extenders from Cisco & Juniper - How they are similar or Different.
Products available under this category in both the OEMs are -
Cisco - Nexus 2000 Switch (FEX)
Juniper - QFX3500 Node
Since I am a Cisco Fan Boy will talk about Cisco First -
Cisco Nexus 2000 is a fabric extender for Nexus 7000 & 5000 series of switches. 2k comes in multiple flavors -
So, You get the liberty to choose whatever suits your requirement. A small list of benefits using Nexus 2000 -
- Single point of management: The Cisco Nexus 2000
Series Fabric Extenders are remote line cards for a Cisco Nexus parent
switch. All device configurations are managed on the Cisco Nexus parent
switch, and configuration information is downloaded to the Cisco Nexus
2000 Series Fabric Extender using in-band communication.
- Software maintenance simplification: The Cisco
Nexus 2000 Series software is embedded in the Cisco Nexus parent switch
software. The fabric extender is a plug-and-play device that
automatically downloads the software image from the Cisco Nexus parent
switch in the same way that a line card downloads software from the
supervisor engine in a modular chassis.
- Switch feature consistency:(make a note of this) The Cisco Nexus 2000 Series forwards all traffic to the
parent Cisco Nexus switch over 10 Gigabit Ethernet fabric uplinks.
Passing all traffic to the parent switch allows traffic to be shaped
according to policies established on the parent Cisco Nexus switch with a
single point of management.
- Effective bandwidth utilization: Through the
virtual PortChannel (vPC) feature support on the Cisco Nexus 5000
Series, a server can be dually connected to a pair of fabric extenders,
or each fabric extender can be connected to a pair of Cisco Nexus 5000
Series Switches, thus giving customers both server and fabric extender
connectivity redundancy and providing active-active connectivity with
twice the bandwidth utilization.
- Reduced power and cooling: Cost-effective 10
Gigabit Ethernet solutions, optimal cabling, device consolidation,
rack-space reduction, and efficient bandwidth utilization all contribute
to a significant reduction in power and cooling needs in the data
center.
- Hardware forwarding Rate: 560 Gbps or 595 mpps
- Over Subscription : 4:1 (Talking about 10G switches only)
Now, Its time to talk about Juniper QFabric QFX3500 -
Trust me friends I wanted to write a lot about this switch but then I didn't wanted to bore you guys by repeating everything said above, but still we will talk about almost everything which would make a difference. lets start -
Juniper QFX3500 switch has everything mentioned above plus it has its own added advantages. Remember I asked you guys to make a note of a point above - "Switch feature consistency"
Now this is the biggest diffrentiator feature. Nexus 2K sends all traffic to the parent switch (7K or 5K) which eventually makes it a 2-Tier architecture. This induces unnecessary delay into the packet and traffic for two adjoining rack has to pass through the core switch, whereas with QFX3500 the scenario is flipped. QFX3500 is a brainy device, It can act as a remote linecard and a full fledged L2/L3 Switch on the same time. It can download the configuration and other informations from the Core Switch (QFabric Interconnect) and deal with the traffic using its local intelligence. Hence it doesn't have to send every packet to the parent switch, which actually provides you fully meshed any to any connectivity.
Specifications of QFX3500 are as below -
- 1 RU device
- 48 SFP+ ports: 36 10GbE and 12 dual-mode 10GbE or 2/4/8 Gbps FC
- 4 QSFP+ ports: 4x 10Gbe or 40Gbps
- Redundant AC power supplies
- Runs Junos Operating System
- Hardware forwarding Rate: 1.28 Tbps or 960 mpps
- Over Subscription : Was unable to find information on this but I am sure it would be very low when compared to Cisco nexus 2K (Will update this soon after getting facts & Figures from Juniper's Team)
Looking at the
QFabric architecture, one has to wonder why Juniper
claims it’s a 1-tier architecture. Honestly, it’s as much 1-tier as
every MPLS/VPN network I’ve ever seen. However, like with MPLS/VPN,
there’s a trick – QFabric uses single-lookup forwarding.
The ingress QF/Node performs full L2/L3 lookup (including ACL checks)
and decides how to forward the packet to the egress QF/Node. The
QF/Interconnect uses the proprietary frame forwarding information to get
the user data to the egress QF/Node. The frame forwarding information
likely includes enough details to allow the egress QF/Node to forward
the frame to the output port.
The expensive part of the user frame/packet lookup is thus performed
only once (whereas you’d get three full lookups in a traditional data
center design using similar hardware architecture). Net result: 5
microsecond forwarding latency across the fabric. Not bad, considering
that the QF/Interconnect itself has three hops.
Conclusion - Do I still need to say this ?