Data Centers Need Big Data Network Analytics, But as SaaS

Cisco recently announced a new data center analytics platform, called Cisco Tetration Analytics, that is designed to resonate with operations teams at medium and large data centers by delivering pervasive real-time visibility across all aspects of data center traffic and activity. The announcement is significant for Cisco in part because it creates expectations of data center functionality to which competitors such as HP and Dell/EMC will likely have to respond.

Cisco's CEO Chuck Robbins underscored the importance of the new platform to Cisco by writing a blog post to highlight the announcement. Robbins said that “data centers can be thought of as the brain of a company, where the most critical information and applications operate and run,” but that unfortunately “we don't know what's happening inside our data centers.”

In its initial release, the Cisco Tetration Analytics platform consists of 36 servers (1RU Cisco UCS C220 Rack Servers) and three Cisco Nexus 9372PQ Switches. Also included will be software sensors that can be deployed on servers to collect telemetry information, and hardware sensors that reside on Cisco's 9000 Series Switches.

To ease adoption, during the initial launch period the platform comes bundled at no extra cost with Tetration Analytics Quick-Start Implementation Services, which Cisco says will “help each customer integrate Cisco Tetration Analytics into the customer's data center, define the most relevant use cases, and transform the data center operations to make them more efficient and secure.” Cisco hasn't yet announced pricing for the platform, but given the amount of equipment that it requires, I have to assume that it will be very expensive.

As management data increases, legacy approaches will be even less able to provide the required visibility.

I thoroughly buy into the need for a Big Data-based approach to data center management. In a recent post I discussed some of the factors driving that need. One of those drivers is that current approaches to network management don't provide the necessary visibility, largely because they enable only a small amount of management data to be stored for more than a brief period of time. Going forward, global IP traffic is expected to almost triple in volume between 2014 and 2019, driving a correspondingly dramatic increase in the volume of network management data. As management data increases, legacy approaches will be capable of storing an even smaller percentage of it, and will be even less able to provide the required visibility.

Beyond storage, we can see from the big iron that Cisco deploys for Tetration that to rapidly pore through all of that network data, and get at the valuable details needed to make timely decisions, you need a lot of processing power.

For most organizations, Big Data network analytics will look more like SaaS and less like a hardware stack.

Because I see the need for a Big Data-based management platform and I believe that some organizations will want this platform to be on site, I think that Cisco Tetration Analytics is an important step in the right direction. To appeal to operations teams at the majority of medium and large data centers, however, what's needed is a Big Data platform that supports equipment from all vendors, offers highly scalable processing, doesn't come with a large up-front cost, and doesn't require third-party services to configure and use. That all points to a platform that is based in the cloud. So while Tetration Analytics is indicative of the industry's direction, for most organizations the fulfillment of the Big Data network analytics promise will look more like SaaS and less like a hardware stack.

Jim Metzler is an independent industry analyst and consultant with a broad background in high-speed data services, network hardware, and network services, including positions as a network manager, product manager, engineering manager, and software engineer.