Cisco has been quietly watching the world around it change, or more specifically the data centre world. What was once a hardware dominated entity is now being touted as a software defined entity.
For Cisco, this must be painful because the shouting is coming from no one other than some of its own partners, specifically EMC and VMware. Cisco quietly watched on the sidelines when VMware acquired Nicira and started touting software-defined networking as the way to go. Recently at VMworld when VMware launched NSX, the launch partners did not include Cisco.
VMworld was about software-defined data centre - which was a covert reference to the end of a hardware dominated world. On the EMC side it is all about software-defined storage (VMware is still in the process of figuring out what exactly it software-defined storage means to them, as was evident at VMworld. For one thing it does not mean the same to VMware and EMC). Between these two partners crowing their own software-defined components and promising to change the data centre entirely, Cisco I believe must have reached a boiling point somewhere in their HQ.
It should come as no surprise (or may be to some it will) that Cisco also has a massive stealth investment operation. Nexus, Andiamo, Kalpana and many others have been absorbed back into the mothership as successful exits. The fact that Cisco was a silent investor in Whiptail was perhaps not widely known. I believe Cisco exercised their option to acquire Whiptail in an act of despair:
- As a maker of a compute and networking platforms, Cisco really lacked the storage layer: Cisco's UCS is a fantastic compute platform. In fact the only reason why NetApp and EMC dominate the integrated systems market, it is because of UCS. However the biggest deficiency with UCS (and really Cisco for that matter) is that while it has an excellent compute (UCS) and networking (Nexus) stack, it really has no built-in persistent storage capabilities. Cisco has to rely on external partners for that capability.
- The industry is heading for scale-out shared nothing architectures and Cisco wants to play in that world: Every storage and systems player is moving their solutions to be scale-out in nature - and many of them are taking it two steps further and adding internal storage as the means to create a shared nothing storage systems - that can also in some cases host compute capabilities locally. This is the approach taken by both commercial and open-source/community based platforms like Hadoop, OpenStack, Ceph etc. This is a lucrative and promising segment and as a maker of a compute platform, Cisco simply could not pass it - or certainly not let its partners (with other self-serving interests) dictate how to go about architecting such solutions.
- The market is becoming multi-hypervisor and cloud-driven: In other words, there are other options besides VMware. With a complete compute-storage-networking layer, Cisco can appeal to a broader ecosystem - Citrix, Microsoft, OpenStack, Red Hat, Oracle - and not be beholden to VMware as a hypervisor platform. Armed with a complete C-N-S stack, they can cast their net far and wide.
So what was Cisco thinking?
- The all-flash array market is heading for a speedy exit and acquisitions will become more and more expensive (eventually): IBM acquired Texas Memory. Western Digital acquired Virident. Pure storage got funded for an IPO. Violin filed for IPO. You get the idea. Basically if Cisco waited long, it would have had to pay the premium for all flash array (or even a hybrid array).
- Acquiring incumbents or well-known start-ups is expensive: If you examine standalone storage suppliers, there is either NetApp or Nimble Storage or a host of other smaller start-ups. Acquiring NetApp or even Nimble would have been a very expensive proposition - and would have not given Cisco what they wanted in terms of the ability to bolster the capabilities of UCS.
- Partnerships will always be an option: The fact that Cisco will embed (some of) the storage capabilities within UCS does not mean that partners will disappear. After all, when they went about acquiring Cisco's competitors, Cisco did not alienate them. But by acquiring a storage stack relatively cheaply, it now has the ability to call the shots - or at least offer options to its clients.
So there you go. Cisco took the plunge and soon will be a considered to be a storage player in addition to be a powerhouse in computing, integrated systems and of course networking.
However Cisco still has some ways to go before it can match the prowess of EMC and NetApp - if that is indeed where it intends to go. For one thing, Whiptail was a start up with limited storage management capabilities in its arrays. Word is it also relied on certain OEM relationships for offering these data management capabilities.
The Whiptail stack is certainly no match for the well-established platforms from EMC, NetApp and even those from IBM, HP and Dell. In fact even in the all-flash and hybrid array space, Whiptail was not quite known for the abilities of its arrays. But the important point is that Cisco has taken the plunge and begun the journey. It should signal two things to its competitors and partners: Don't take Cisco for granted and don't think they will be shy to encroach on your turf if you encroach on theirs.
It was ironic that Cisco's announcement came just a week after EMC announced a flash-optimised version of the VNX, that leveraging MCx or multi-core operating system. EMC also went to great lengths to tout an all-flash VNX2 and the success of the VNX2 in integrated systems, notably VBlock and VSPEX. So the biggest question really is what is the future of Cisco's VCE partnership with EMC and VMware?
It needs to be said that the unforeseen casualty of Cisco's move could be on their relationship with NetApp. Poor NetApp - they can't seem to catch a break. They had hit quite the stride with FlexPod. I wonder how much of the FlexPods will now be serviced by the Cisco UCS-Whiptail combination? Will Cisco stay committed to NetApp (who unlike EMC and VMware has stayed committed to Cisco)? Or will Cisco simply bite the bullet and acquire NetApp?
Posted by Ashish Nadkarni