What is the aim of HP Virtual Connect Flex Fabric?
- Simplified wire-once connection to LANs & SANs Then move workloads or add, move, or replace servers transparently to LANs & SANs in minutes, not days
- Converged up to 95% fewer NICs, HBAs and switches at the server edge, up to 65% less cost, 40% less power and cooling costs
- Integrated by working with any Ethernet or Fibre Channel network – HP Networking, Brocade, and Cisco Catalyst/MDS/Nexus – Integrated on HP G7 server blades
Over time, customers have converged application workloads onto fewer servers. While virtual machines can trim your server requirements, they tend to bloat your network infrastructure – requiring more expansion cards, switches and cables. Traditionally networking at the server edge tends to become complex, inflexible, and costly. Customers want to get as close to “wire-once” as possible. The thing that stands in the way is that every time they buy and add a new server later, the LAN and SAN people need to get involved again to help the system administrator because when you add a new server, you’re adding new connections to the networks and they have to adjust.
Before we go any further, let’s review a couple of definitions. First, network convergence refers to the use of a single network to transmit both LAN and SAN traffic. iSCSI is an example of this, however, most of the networked storage in data centres is fibre channel.
- To converge LAN and FC storage, two standards are under development, FCoE and CEE.
- FCoE simply encapsulates FC frames so they can be sent over an Ethernet network.
- But that is not even half of the prerequisite for converging today’s LAN and SAN infrastructure. In addition, Ethernet must be enhanced to make it reliable enough for storage traffic. FC storage cannot tolerate dropped packets in congested network situations. So, another standard called CEE is being developed to make Ethernet more reliable … or “lossless.” It’s turning a sometimes rough road into a smooth highway.
- Since the point of converging networks is to simplify your data centre, it makes sense to converge first where you get the most bang for the buck. Most customers tell us that’s at the edge where the servers meet the networks, where both the majority of ports, cables and traffic reside. In typical data centres, most of the cables connect servers to edge switches, most of the switches are connected to servers, most of the data traffic is from one server to another nearby, and most workloads share one edge.
- The rest of the data centre contains architecturally complex equipment, but today there is relatively less opportunity to simplify infrastructure. Port-P believes that focusing attention on simplifying the server edge, will yield the most benefit while converging networks in the distribution and core layers is not as cost-effective and based on standards maturity, is most likely premature.
- So, network convergence is a topic of considerable interest to all of us in the IT industry today. As with any emerging technology, IT planners and decision-makers are challenged with determining the adoption strategy that’s right for their business.
- The promise of convergence is that it will reduce the cost of your LAN and SAN networks in terms of equipment to qualify, buy, to power and cool, to maintain, and to manage
However, there are risks with convergence that can compromise those benefits for your business. Depending upon the extent of your convergence plans, you may find challenges in your network governance, performance, and infrastructure.
– When you’ve combined two independent networks with previously independent teams managing them, who is responsible for the combined operation and SLAs?
– If you collapse multiple networks down to one set of switches, cables, etc. how are you going to make sure that it is sized to handle all of the loads and priorities?
– And, the new standards for the converged infrastructure will require new equipment. That’s easy to do in a new data centre, but if you’re modifying or adding onto an existing one, that transition may be challenging.
- If Convergence is going to save you enough money, then you’ll need to take these issues into account and plan for them. And, you may find that network convergence in one area of your data centre is a good idea, but not worth the investment in another area.
HP FlexFabric is a highly scalable data centre fabric architecture.
- Virtual Connect uses Flex-10 to simplify hardware on server blades and interconnects.
- The same 4:1 consolidation now includes storage so you can eliminate HBAs and Fibre Channel switches in addition to the NIC and Ethernet switch reduction of the initial implementation of Flex-10
- Whether you are using integrated or mezzanine FlexFabric Adapters, you need only 2VC FlexFabric interconnect modules. And when using mezzanine cards, you are no longer required to have primary VCM modules in bays 1 and 2
HP Flex-10 delivers unprecedented simplification of server connections to LAN and now SAN
Eliminate Adapter Cards, Interconnects and Cables
HP’s unique approach to integration and convergence eliminates more networking components than any other solution available. VC FlexFabric modules and adapters eliminate a lot of hardware. For example, you could wrap the outside perimeter of the 2010 World Cup stadium in Dunham South Africa more than 70 times with the cable savings versus a traditional rack mount data centre of 3000 servers
For more information on how FlexFabric could save you money without disrupting your current IT practices whilst maintaining investment protection with built-in flexibility today and in the future, please speak with one of our server storage specialists or email essn@port-p.com.
