Upcoming Events

HDI Service Management 2010 Conference & Expo
October 6-8, Miami

IT service and technical support professionals gather at the annual HDI Service Management Conference & Expo to explore some of the hottest topics affecting IT service management. The half-day conference workshops provide the processes, frameworks, templates, and tools to help you meet the service demands of your business..

More Events »

Subscribe to Newsletter

  • Keep up with all of the latest news and analysis on the fast-moving IT industry with Network Computing newsletters.
Sign Up

Cisco Overlay Network Bridges Distributed Data Centers

Tags:

Channel: Data Center, Networking & Mgmt, Virtualization, WAN & App Acceleration

Cisco continues its march to complete its Data Center 3.0 vision with a proprietary method to interconnect remote data centers called Overlay Transport Virtualization (OTV), extending a layer 2 Ethernet network over a WAN. The claim is that OTV is simpler. Cisco also counted 10Gb Base-T Ethernet modules for the Catalyst 6500 and 4900 switches, and new IO modules for the Nexus 7000. Additionally, they are furthering their Wide Area Application Services (WAAS) product with VMware as well as automatic recognition, acceleration and optimization of SaaS services.

Organizations with multiple data centers typically face expensive and time-consuming work when they want to bring new services on-line or change existing services. The time to get a new circuit up and running, whether it's Carrier Ethernet or MPLS, can be weeks or months. Many WAN service plan changes incur additional costs, and that's just to get the network in place. Then, services have to be enabled. Cisco OTV is an overlay network that the company claims can be provisioned in minutes. Since OTV is an overlay on top of IP, the traffic can be routed transparently over any IP-based WAN connection. OTV is built to have low-management overhead, replication, is fault-tolerant and includes multi-path, optimized paths between nodes. OTV is available on Nexus 7000 in April, 2010 through a software upgrade.

It's a bridge after all!

The Curmudgeon
Cisco spent all these years trying to make the network smarter only to turn the Internet into a bridge? Guess that's why they can't let go of that logo.


Short on details, OTV uses a proprietary protocol to encapsulate Ethernet frames in IP at the network edge and transports them to the remote data center where the Ethernet frame is decapsulated. OTV extends a layer 2 network over the WAN. Broad control is built into OTV so that unnecessary broadcast traffic, including broadcast storms, doesn't traverse the WAN while allowing applications that require layer 2 connectivity, such as VMware's VMotion, to work.  The protocol allows for each Nexus 7000 to exchange control planes and divide MAC address knowledge among each peer automatically.

The downside of overlays is the additional protocol encapsulation overlay cost, in this case, typically 40 bytes per packet, which can add up fast. Also some applications, like storage protocols, may not behave well on an overlay network, where WAN characteristics, like delay and jitter, can vary. For example, carrier Ethernet WAN deployments that are used for high-capacity WAN connections tend to have more granular specifications for delay and jitter, while MPLS and IP WANs can vary widely.

In addition, since OTV is not encrypted, it is unlikely you will want to use this over an untrusted WAN. OTV packets may also pose a problem for firewalls and other network equipment that processes IP packets. Early implementations of IP VPN traffic, such as IPsec and PPTP, both of which encapsulate IP traffic, had issues with firewalls that couldn't process the packets properly. Whether OTV will suffer more limitations will be borne out in deployments.

Page:   1   2  Next  »

Add Your Comment:

Premium Content

Don't Stop At VoIP
June 2010

Network Computing June 2010


Salary

Video