VMware has just announced that it intends to acquire the SD-WAN company VeloCloud in an attempt to expand the capabilities of its networking business unit. No one can deny that VMware NSX is a very capable networking product, but it does have gaps and is incomplete. In reality, it can only create software-defined network islands surrounded by legacy physical oceans. This has always been a yawning gap in the product; it can only network virtual devices. That said, in VMware’s world, there is no such thing as a physical server. However, on planet Earth, there are still many physical devices attached to corporate networks.
On paper, this seems like a sensible acquisition. VMware is not in the SD-WAN space, and VeloCloud has over a thousand customers. Among them are some of the giants of the telco world, including Sprint Corporation, AT&T, and Deutsche Telekom. The acquisition does expand VMware capability and moves it out from the traditional data-center space, allowing it to control the flow of traffic between sites.
It will also elevate its story, allowing it to compete better against the usual suspects of Cisco and a newly invigorated Extreme Networks.
The real question is why VeloCloud and not another of the SD-WAN companies out there. The cynic in me says it came down to price. VMware is not known for paying over the odds for a product (Nicira an obvious exception).
VeloCloud’s product is actually quite a decent SD-WAN. It is composed of three core components, these being:
- A network of service gateways deployed at cloud data centers around the world to provide redundancy, scalability, and on-demand agility
- Enterprise-class appliances that facilitate connectivity to private, public, and hybrid traffic flows coupled with the ability to host NFV services
- And finally, an orchestrator that provides a centralized and automated config, provision, and monitoring solution across the whole network, from branch to cloud and data center.
What is more important is the ready-made customer base of over one thousand. Remember, these are not insignificant customers, but include major telcos and enterprises.
A more important question to ask is why VMware is acquiring the company. Is there synergy? According to VMware EVP Shekar Ayyar, who has responsibility for strategy and corporate development and is also the GM of the Telco NFV group. There, too, as outlined previously, VMware has three key tenets that are taken into consideration when making an acquisition decision. These are platform acquisitions, industry acceleration acquisitions, and finally, technology tuck-in, but what do these actually mean? And how does this relate to the VeloCloud purchase?
Well, on paper, it actually ticks all three boxes. Platform Acquisition—“check”: a VeloCloud purchase moves VMware into a new market, that of wide area networking, and strengthens its position with telcos and global corporations. Industry acceleration acquisition—“check”: the acquisition of VeloCloud gives VMware a ready answer to any WAN-based questions that have previously been difficult to answer, trying to force a round-peg NSX into a square-shaped WAN hole. To be fair to VMware, it would have needed to partner with a company like VeloCloud to correctly fill that gap. And finally, a technology tuck-in “check”: there is no doubt that NSX is a capable product; however, there are significant gaps in capability and weaknesses in its arsenal that its major competitor Cisco could easily answer with its SD-WAN acquisitions Meraki and Viptela and its IWAN product. Any company that moves beyond a single site—and in today’s multicloud world, this is almost every company that has a WAN. Some people call that a internet connection, but it is still outside your direct control. Solutions like MPLS and dark fibre connections are costly and time consuming to acquire and deploy. SD-WAN can simplify these connections with a policy driven and automated orchestration layer (northbound RESTful APIs) to provide flexibility and agility.
Of course SD-WAN, like SDN, is not for everybody. Then again, neither was virtualization in the beginning. Only those large enterprise companies could really gain any savings, right? And there lies the rub and the problem that companies like VMware and the other software-defined companies have to answer. With the introduction of ESX technologies, there was a visible savings. I was not running several server workloads on a single physical server. There were genuine verifiable savings that everybody could see. SDN savings are much more nebulous. I mean, I still need the same number of ports to physically connect my hosts and other assets, too. Concepts like separate control layer and network function virtualization are confusing to many, but there is no doubt that the march of software-defined networking is gathering speed
But this is missing the point of true SDN. Yes, I do need those physical connections, but they can now be simple devices. The magic is now provided by software (yes, it was always provided by software, but that software was decentralized in individual switches, routers, firewalls, IDS/IPS devices, etc. Each moving piece had to be individually managed. It was like a many-headed monster, requiring constant love and attention to do its job. At any time, any single head could throw a tantrum and bring down the network. To add insult to injury, none of the individual heads really spoke the same language. and they. only just tolerated each other. SDN is more like a multicelled organism with a central nervous system controlled by a single brain. I just hope that the integration of VeloCloud into the NSX portfolio does not give it a split personality.