Last week I attended a pre-briefing on the HP FlexNetwork architecture launch. Like other major equipment vendors, HP has structured its networking offering into an end-to-end architecture. FlexNetwork has three major components: FlexFabric for the data center, FlexCampus and FlexBranch. These can be managed together with unified FlexManagement.
We all know that a great marketecture does not a good network make. So let's take a look at the real technology behind the barrage of mandatory buzzwords used to describe the FlexNetwork components (e.g., open, scalable, secure and agile).
For the campus, HP launched a new series of high-end chassis switches. The A10500 series offers up to 128 10 Gigabit Ethernet (GbE) or 384 GbE ports in a single chassis. You can stack up to four chassis in a single virtual box using HP’s proprietary IRF technology, totaling more than 1000 GbE ports. (It is important to note, however, that this capability is limited to the new campus switch and does not extend to HP’s data center framework.) Performance figures for the A10500 HP switches are good. It is puzzling that HP chose at launch to compare the brand new box to the decade old Catalyst 6509.
On the downside, the A10500 is no more than a high-speed campus switch. It has no advanced features that would make it suitable for smaller data centers (such as support for Transparent Interconnect of Lots of Links (TRILL)/Shortest Path Bridging (SPB), edge virtual bridging (EVB)/Virtual Ethernet Port Aggregation (VEPA) or data center bridging.
In the data center network realm, HP has finally published a data center architecture design framework. The FlexFabric Reference Architecture is an excellent and badly needed step that HP had to take, but still a far cry from Cisco’s detailed design documents. But here again, the reference architectures (spanning everything from a single-tier design with blade servers to a large-scale, three-tier design) are low on emerging data center features.
The "Key Data Center Networking Requirement" part of the document mentions all the necessary technologies, from TRILL/SPB and DCB (including QCN) to EVB/VEPA. The "HP Data Center Reference Architecture" part of the document, however, mentions only FCoE, which is immediately split into Fibre Channel and 10 GbE in the FlexFabric, or top-of-rack, switch.
From a technology standpoint, a more interesting part of the FlexFabric story is the TippingPoint S6100N intrusion detection system (IDS), a standalone appliance that can insert itself between a virtual machine and the rest of the network using VMware’s DVFilter application program interface. Combined with tight vCenter integration, the S6100N IDS can provide protection to a virtual machine even after it has been moved to a different physical host via vMotion.
Not surprisingly, network management seems to be the real strong point of HP’s networking. During last year’s Gestalt IT Net Field Day, delegates were impressed by the Intelligent Management Center (IMC) demo, and HP continues to make the product better. The latest release adds vSwitch and VM monitoring, making IMC a product that can manage physical and virtual parts of your network, regardless of the vendor used to build the networking infrastructure. (IMC supports more than 1,000 different devices from Cisco.)
It seems that HP continues to make gradual improvements and offer higher performance for a lower price. It’s probably the right approach for the company's customer base, but it definitely doesn’t excite my inner geek.
About the author: Ivan Pepelnjak, CCIE No. 1354, is a 25-year veteran of the networking industry. He has more than 10 years of experience in designing, installing, troubleshooting and operating large service provider and enterprise WAN and LAN networks and is currently chief technology advisor at NIL Data Communications, focusing on advanced IP-based networks and Web technologies. His books include MPLS and VPN Architectures and EIGRP Network Design. Check out his IOS Hints blog.