Summary Bullets:
• Engineers have been seeking for decades to instrument and monitor the data center in real time with varying, but always limited, degrees of success.
• With Cisco Tetration Analytics platform, the company hopes to take a leap forward with a witch’s brew of big data, machine learning and IoT-scale instrumentation that together promises to bury traditional network monitoring under a mountain of analytical insight.
I remember the ‘80s fondly, mostly because I’m a product of ‘the 80s, an era I think we can all agree was a lot more fun than our current epoch, the, ah, naughties. The clothes, the music, the art, the attitude, but mostly the hair, they all spelled out in big bold, color the idea of optimism. Optimism that we could do some seriously cool stuff. Take Robert Zemeckis’ 1985 pseudo-dystopian dose of frenetic energy, Back to the Future. As a kid, I just “knew” that time traveling cars could happen and that there was no reason at all why we couldn’t do just that — given enough plutonium of course.
Obviously that hasn’t happened (yet). And let’s not even discuss the existence of hoverboards (besides, that’s an entirely different movie). But I swear, after listening to Cisco talk about its new Tetration Analytics platform during Cisco Live! Last week, I can confirm that my ‘80s optimism is alive and well. I believe that Cisco can actually create a future utopia of “The Cloud Datacenter,” as referenced by Cisco itself, where in real-time, every packet header from every source (cloud, app, server, switch, etc.) can come together and mean something.
We don’t know yet whether or not Cisco can pull off such a bold move. Frankly, it seems easier on the surface to construct a time-displacing DeLorean. Engineers have been seeking for decades to instrument and monitor the data center in real time with varying, but always limited, degrees of success. Is this the answer? Was SOA the answer, or SDN for that matter? Of course not, at least not in terms of each defining a top-down “idea” that needed to be forcibly imposed upon IT in order to yield any value. I don’t believe a single architecture can ever solve a real problem like real-time policy compliance modeling and enforcement. There needs to be a product.
And that’s exactly what Cisco has done here with Tetration Analytics. Contrary to its nature (Cisco loves frameworks and reference architectures), the company has built actual software and furnished actual equipment (select Cisco UCS servers and Nexus 9000 series switches for starters) with that software. And I might add that it is built on top of some pretty well known open source big data projects including Kafka, Druid and of course Spark. But that’s not why I think Cisco will succeed with Tetration Analytics. It all comes down to smarts, artificial smarts to be more specific.
Cisco Tetration Analytics platform makes great use of some ideas Cisco has been tooling around with for some time now and concerning self-learning networks (mostly down to Cisco Fellow JP Vasseur). Tetration applies machine learning and mixes that with behavior analysis to not only provide visibility across everything in the data center, but also to automatically build and maintain a working understanding of the application dependencies, interactions, and requirements of the solutions running across the wires, switches and servers within that data center.
With such knowledge, IT managers can, as just one example, automatically generate and enforce a comprehensive and constantly evolving whitelist of applications (blocking all, allowing all, allowing some), while significantly improving upon traditional blacklist approaches to security. This sounds promising, even futuristic, to be sure. As with all futuristic endeavors, and as with the famed, fictional DeLorean of Back to the Future, Cisco Tetration Analytics platform is a one-off, a lark even, predominantly specific to Cisco’s in-house technology. But without such examples of what’s possible, how else are we to achieve the impossible?
IT Connection 