Don’t you love it when new firmware ships for your iPhone or Android device? New features and a great price—FREE! The bad part? Downtime. While updating your device firmware, you’re down hard; Five minutes maybe - No calls, no email, no tasks, nothing. Just you, the little spinning dial, and what seems like an eternity. When it’s done—oh the joy—it’s like a new device. Upgrading to UCS 2.0 is sort of like that, except the whole downtime part—the process is non-disruptive.
Announced at CiscoLive in July, UCS 2.0 has finally arrived. Plenty has been written about the new hardware and the benefits and flexibility to be had in both scalability and flexibility. Combining the new fabric interconnect (6248UP) with the new fabric extender (2208) and new VIC (1280) is great in big deployments with high throughput requirements and hundreds or thousands of VMs. But for the vast majority of my customers, the real news here is the software—interoperable with every UCS deployment out there, easy to upgrade without downtime, and free. Add in the fact that you can manage the whole upgrade process for a farm of up to 160 servers from a single pane of glass, well, you’ll be loving life! UCS! UCS! UCS!
Two of the biggest features available with UCS 2.0 (and completely relevant to us as recently as this week) are iSCSI boot from SAN and configuring disjoint Layer 2 networks. Each capability addresses limitations we had to previously work around when deploying UCS in our customer networks.
Since the beginning, UCS has supported boot from SAN via FC or FCoE and that works great if you have that environment. But what if you don’t and you want to be completely stateless and boot ESXi from an iSCSI target on your SAN? Before UCS 2.0, you would be out of luck. Sure it’s not THAT big a deal—but if you want to squeeze every bit of benefit out of UCS—you really want to have everything offloaded but compute and memory. It makes it so much easier to upgrade, perform maintenance, provision new systems, and respond to outages. Now, with UCS 2.0, you are good to go.
Configuring disjoint Layer 2 networks—hmmm. Try to explain that one to your mom. Or try to explain to your customer why they can’t hook their servers or VMs into multiple disparate physical networks. You won’t have to anymore, as now each vNIC can be pinned to a specific uplink and each uplink to a specific network. In the case of one of our customers, their existing infrastructure was built with a separate backup network in order to keep that traffic off their core switch and limit the possibility of impacting performance on production systems. Our best option with UCS 1.4, was to route to the upstream core switch and push traffic to the backup network, essentially eliminating the benefit of the intended design. With UCS 2.0, that is no longer an issue. This feature is also great in multi-tenant environments too!
All in all, UCS 2.0 brings a lot to the table. Kudos to Cisco for continuing to grow an already great platform—and especially for not leaving early adopters out in the cold. UCS! UCS! UCS!