Like many other attendees at Cisco Live this year, I was stunned at the Nexus 7700 announcement. When the vendor's reps read the stats about the new device, 384×40 Gigabit Ethernet (GbE) or 192×100 GbE ports, 83 Tbps total switch capacity, it took some cycles to absorb.
Admins now face a cacophony of protocols, standards, optimization frameworks, security concerns, service models and more on ever more concentrated links.
Why on earth would anyone build something like that? It's just a monster. Later, after a hundred other networking conversations, I thought back to 20 years of splashy hardware premiers and how antiquated the bulk of those devices are today. And in that context, especially considering the ever-wider adoption of fabric, the development of the Nexus 7700 is nothing more than a natural evolution -- even if it seems a moonshot today. The simple truth is that the accelerating growth of applications and data on our networks and the associated management challenge is nothing short of astounding.
You kids get off my lawn
Once upon a time I was at a trade show where a vendor (I think it might have been Intel), was showcasing a very early preview of a then super astounding 10 GbE network interface card (NIC). The demo had a big spool of fiber on a shaker table, wobbling and bending foot-long leads into matched NICs. An array of HP oscilloscopes displayed real-time transfer speed and error-correction data proving that 1) the motion was inducing error and 2) the NICs were sorting it out, resulting in a smooth transfer rate. At the risk of sounding older than I am, this was pretty amazing at the time and it was difficult to imagine what anyone could do with 10G links, let alone afford them. My company’s 10T Ethernet conversion of Token Ring cabling, even using ridiculous, ferret-sized IBM hermaphroditic Type 1 to RJ45 dongles, was working just fine, thank you very much.
Today it's difficult to believe that was ever the case. Capacity has increased so much, by so many orders of magnitude, in such a short time that it seems like we've always had affordable, very high capacity media, routers and switches. What was once a novel and probably expensive one-off trade show demo is now a regular admin's chore in a closet with a well-worn optical time-domain reflectometer (OTDR) tester. Meanwhile, that stack of scopes has been condensed into a set of durable OTDRs that manage optical media nearly as easily as copper. If you're planning even a medium campus LAN, you've looked at pricing and are likely to use 10 GbE to connect your distribution and access switches -- typically dual 10 Gb links with volume shadow copy service for chassis redundancy -- knitted together with EtherChannel. In that context, the Nexus 7700's ports, with their support of 40 GbE and 100 GbE, further drive down the cost of 10G links.
Fewer parts, more services to monitor
The advent of higher bandwidth links also means changes to network performance monitoring systems. The challenge is not that these systems have more devices to connect to; in fact, they often allow network planners to decrease the number of boxes. The real challenge is this: Ever-fatter pipes, combined with more skilled network admins, have been an invitation for businesses to put every service they can think of on our networks. As a result, monitoring tools have grown to include once-exotic features right out of the box.
Admins now face a cacophony of protocols, standards, optimization frameworks, security concerns, service models and more on ever more concentrated links. Gone are the days when file sharing was usually the biggest bandwidth hog on access switch ports. It's not enough to simply report bandwidth utilization, CPU and memory, or to provide up/down status. That's because high-speed networks can be effectively broken even with all the lights green. Route flapping detection, automated topology mapping, dependency detection, VSAN monitoring, modular chassis metrics, Unified Computing System and more are everyday concerns on these networks. Network admins upgrading to 10 GbE and beyond now demand all these features and more to effectively manage ever-higher throughput links.
Perhaps new hardware announcements aren't really a measure of age, but markers of evolution. They're simply a byproduct of advances that keep admins feeling young and engaged, even as years config the first few follicle interfaces to gray. The Nexus 7700 doesn't make me feel old, just really experienced.
About the author:
Patrick Hubbard is a head geek and senior technical product marketing manager at SolarWinds. With 20 years of technical expertise and IT customer perspective, his networking management experience includes work with campus, data center, storage networks, VoIP and virtualization, with a focus on application and service delivery in both Fortune 500 companies and startups in high tech, transportation, financial services and telecom industries. He can be reached at Patrick.Hubbard@solarwinds.com.
This was first published in August 2013