I met the lovely folks from Tail-F at Networking Field Day 7 earlier this year, and I was impressed with their NCS product. If you haven’t already seen what Tail-F is up to, do take a moment to read my post from last month and maybe even view the videos I’ve linked to so that you can hear their presentation directly.
This week brought exciting news for the Swedish firm with the announcement of Cisco’s intent to buy them for $175M. First off, let me say that despite the salt licorice they brought to NFD7 (it’s a taste I tried to like, but just could not adjust to) , I am delighted to see a company succeed like this and, presumably, finally get to cash in on all the hard work they have put in!
I am a little more concerned though about what this means for their products.
Tail-f’s ConfD is a clever service that effectively acts as middleware between a managed object (e.g. a network device) and management interfaces like SNMP, Web UI, CLI, NETCONF and REST. The idea is pretty much that ConfD allows you to deploy any or all of these services at high velocity because of the built-in capabilities of the platform, and the power of the YANG models used to map functionality within the system. Tail-f’s website says:
Tail-f technology is used by leading networking companies – including seven of the ten largest global networking providers – to develop the next generation of network configuration management equipment and software.
They don’t list out their biggest clients by name, but you can make some pretty shrewd guesses as to who some of those seven companies might be. The indirect claim made is that if you’re logging into a network device today, chances are you’re actually logging into ConfD. If I were one of those seven companies (or presumably one of those six companies that are not Cisco), I’d suddenly be very concerned to realize that a key component of my products was just purchased by a competitor. What will the future hold? From the Cisco Press Release:
The acquisition of Tail-f accelerates Cisco’s cloud virtualization strategy of delivering software that increases value to our customers’ applications and services, while supporting Cisco’s long-standing commitment to open standards, architectures, and multi-vendor environments.
I love Cisco, but many people in the networking industry might raise an eyebrow at Cisco claiming a long-standing commitment to open standards given the number of times the word “proprietary” ends up appearing in close proximity to the word “Cisco”. Don’t get me wrong; Cisco has a history of pre-empting standards with their own technologies or creating enhanced features around existing standards that are incredibly useful, but whether you consider that to translate to a commitment to open standards, I don’t know. Perhaps the past is not an indicator of the future. Either way, if I were a ConfD customer, I could not help but be worried that a key part of my product architecture was no longer independently held.
Network Control System (NCS)
The NCS product in many ways extends the ConfD capabilities and brings them right to the end user. From their website:
NCS provides a single network-wide interface to all network devices and all network applications and services, as well as a common modeling language and datastore for both services and devices. A transaction engine handles transactions from the operations at the service layer to the actual deployment of configuration changes in the network
I could not have put it better. The CLIs (Junos style and IOS style) offered by NCS match the same CLI options available in ConfD, coincidentally enough. And with a due sense of irony, when deploying changes using NCS, chances are you will be using a Tail-F product (NCS) to connect to a Tail-F product (ConfD) on network devices in order to implement any changes. NCS again uses YANG data models to map declarative needs to procedural implementation, and the synergy between these two products is very strong. In particular, the multivendor capabilities of NCS are incredibly attractive from a network automation and management perspective. Once more with this acquisition I have to wonder whether this is at risk too, notwithstanding Cisco’s “long-standing commitment to […] multi-vendor environments.”
The biggest game in town for device management has just been bought by the biggest game in town for devices. For Cisco this is a brilliant move, boosting their own capabilities in this area while simultaneously, I would imagine, sending shock waves through their competitors who may now be regretting putting their money on this particular horse. Any hopes that Cisco might just let Tail-F get on with eating salty licorice and doing their own thing are pretty much wiped out in the press release:
“Tail-f employees will join Cisco’s Cloud and Virtualization Group”
Cisco has always been the king of acquisitions, assimilating companies and their employees with a deftness so often missing elsewhere in the industry. The down side to this – depending, I suppose, on your perspective – is that it means Tail-F will be swallowed up very rapidly in to the larger company.
AT&T has been pushing its Domain 2.0 initiative, and recently named Tail-F as one of its Domain 2.0 suppliers, which is a huge deal for Tail-F. And now it’s a huge deal for Cisco too, who by this acquisition are back in the door with AT&T (they aren’t one of the other five named companies). I understand that AT&T’s Domain 2.0 strategy is to try and move towards full Network Functions Virtualization and provision of networking functions on generic x86 or similar “white box” hardware platforms. Tail-F is well placed in this environment to provide the programmability layer that AT&T will be wanting. Since AT&T apparently isn’t buying Cisco hardware for this new architecture – which must be a painful loss for Cisco – getting back in through Tail-F is likely very important to Cisco.
Winner, Winner, Chicken Dinner!
I believe this is a big win for Cisco and a big worry for everybody else. How open Cisco remains in terms of making such a key technology availably to their competitors going forward remains to be seen. I’m doing my best to remain optimistic, but only time will tell.
What do you think?
Hi John – great post and insightful analysis.
I am also wondering what Cisco may do with ConfD. Perhaps they might include this technology in “legacy” products that do not currently offer any programmatic interfaces (i.e. only current option is CLI-scraping/SNMP). If so, then the combination of ConfD and NCS as you state could be quite significant (cross-platform modeling in YANG/etc).
Time will tell. The more “programmatic” networking devices are available the better we as a community will be as we further our journey along SDN/NFV/DevOps/etc.
The Tail-f acquisition completely caught me off guard. I think there were a lot of people out there looking at it because it wasn’t Cisco and it wasn’t Juniper. Nonetheless “major networking vendors” were licensing a UI owned by an outsider. This is a huge win for Cisco, a potential loss for others. The question is where the products go from here and where Tail-f customers will land.