The gravitational pull of Amazon Web Services (AWS) can be detected in almost all major cloud platform or service developments and this is very much the case for the newly forged alliance between Cisco and Google. By that very token It should be good news for broadcasters and operators because it adds a serious counterweight to AWS beyond Microsoft Azure, unless we also count IBM’s and Oracle’s cloud efforts which have gained less traction in the TV sector.
The tie up does make sense on various levels, focusing purely on hybrid cloud deployments which is very much where the money is going both inside and outside the broadcasting sector. Whether hybrid is a temporary migration step while the private component is phased out once public clouds can be fully trusted is almost irrelevant and in any case most enterprises will not have made their minds up on that at this stage. The point is that hybrid cloud is the perfect compromise for most larger operators or broadcasters right now. Public clouds have their roots in earlier SaaS (Software as a Service) offerings while private clouds have their origins in the data center from which they have evolved through virtualization and automation that makes them operate more like public clouds, with separation between control planes and underlying hardware. Hybrid cloud combines the two to modernize the date center and move some functions out into a public cloud such as AWS provides, with virtualization and a more adaptable approach that allows a best of breed buying strategy, which is certainly favored now by many larger service providers.
For video service providers, the attractions of hybrid cloud are both economic and functional. They can retain some critical areas on their own private cloud while taking advantage of the scale economies of a major public cloud service for storage and distribution, at the very least of non-premium content. They can also deploy services like network PVR at much lower cost and potentially with global reach through geographically dispersed operations. Desk top apps can be replaced by SaaS delivery and new products can be brought on much faster.
Hybrid cloud has therefore become the battle zone for the major IT players of all pedigrees, from IBM to Amazon, which has caught out some of the big vendors including notably Cisco and Google. Cisco originally launched its Intercloud in January 2014 which ironically as it turned out was billed as the “cloud of clouds” incorporating new functionality based on an SDN (Software Defined Network) model to optimize movement of workloads between existing established public clouds, including AWS and Microsoft Azure as well as VMware’s offering. SDN separates higher level network control from the underlying mechanics of packet forwarding and routing across any multi-hop IP network.
For Cisco the move into public cloud looked like a no brainer at the time. In the 1990s, it won the IP routing battle to become the dominant vendor of iron for proliferating corporate networks as well as the exploding Internet but then reinvented itself as a provider of networking software and services on top. It also entered new domains that required connectivity, including broadcasting which began with the acquisition of Scientific Atlanta for $6.9 billion in 2005. The rise of the cloud raised a similar threat to Cisco of being isolated as a commodity provider of connectivity, especially as virtualization meant that the big cloud vendors were not buying that many routers as they could run those functions on COTS hardware.
It turned out increasingly that Cisco’s enterprise clients were moving their workloads onto AWS and planned to keep them there, leaving not many crumbs over for InterCloud. Microsoft’s Azure also hoovered up to a rather lesser extent. Intercloud became unviable and in March 2017 the inevitable closure announcement came, effectively writing off $1 billion of investment. Cisco indicated then that it would help its customers move their enterprise data to an unnamed cloud provider, which was clearly AWS, although with support for Azure being added. This inevitably sowed seeds of confusion for customers when they heard of Cisco’s second attempt with Google.
Meanwhile the extraordinary advance of AWS had claimed other scalps. Texas based Rackspace, one of the original cloud services pioneers dating back to the 1990s and promoter of the successful OpenStack open source cloud platform had been one of Amazon’s fiercest competitors, but then surrendered in October 2015 to become an AWS partner, after having flirted with Azure also. It then went private by taking $4.3 billion from Apollo Global Management in August 2016 and is still a going concern but now just a bit player.
VMware was another case of the mighty falling, having initially struck out on its own like Cisco but then threw in the towel by reaching a strategic partnership with AWS in October 2016. Another victim of sorts was storage system vendor EMC, although getting $67 billion through acquisition by Dell in August 2016 hardly looks like a disaster. It did reflect how enterprise customers were not buying so much of its hardware because they were finding it cheaper to store data on AWS, just as SDN and virtualization have been strangling Cisco’s golden goose.
Google found itself in a rather different position as a huge force in search, online advertising and web based services. Although cloud has played an increasingly big part in these offerings, Google was slow to invest seriously in cloud computing as a service itself, lagging well behind AWS and Azure. It lacked an enterprise presence, which was hampering its efforts to get a leg up into the hybrid cloud arena as that became the latest area to proceed from hype to enlightenment as Gartner would put it. Google was therefore in a poor position to capitalize on the mass migration among enterprise customers, including public sector organizations as well as corporations, from traditional data center operations to hybrid cloud.
This is where Cisco came in for Google, with its dominant position in data center networking with a large sales force well embedded into this customer base. However, Cisco was also constrained by having failed to shake off its shackles as the provider of plumbing and so the involvement of Google gave it much more clout within the developer community, as well as with more technically advanced enterprises, through its leading position in Machine Learning and AI, two other buttons of the moment.
Google is promoting AI and ML as part of the Cisco alliance for analytics and monitoring, which it sees as essential for a successful hybrid cloud strategy. It believes that even AWS has left a niche unfilled in achieving a tight and yet flexible integration between the public and private cloud components.
The priority of the partnership, again according to Google, is to develop whatever additional tools are needed to align runtime and production across the public and private cloud environments. This will be built around the two open source platforms that Google and Cisco have singled out as foundations for their strategy, Kubernetes and Istio.
Kubernetes was developed by Google and announced mid 2014 as a set of building blocks comprising tools for deploying, maintaining and scaling inter-operable applications within an open cloud service. The components were designed to be loosely coupled so that they can operate within as wide a variety of workloads as possible and extensible, so they can embrace new functions in future. The extensibility is enabled by the Kubernetes API, which is used by internal components as well as extensions or so-called containers running on top. Kubernetes is based on the idea of the software container as a standard unit for deployment, confining its parts within an interoperable exterior. However Google added a higher level of abstraction called a pod, which is co-located on the host machine and can share resources, with a unique assigned IP address. This helps with scalability and efficiency within a large cloud, ensuring that applications can run and access ports without risk of conflict.
Istio was then designed by Google in partnership with IBM and transportation app provider Lyft as an extension of Kubernettes, designed specifically to assist in development of microservices comprising smaller chunks of software that can be deployed and tested much more quickly. The first alpha release was announced in May 2017, offering developers finer-grained visibility and control over traffic than they had enjoyed before, without having to change any application code. It also includes tools to help enforce security and compliance around microservices.
Google and IBM have described it as a service mesh or substrate for microservice development and maintenance, decoupling operations from the software components of the microservices themselves. This looks like an essential step to make a genuine microservices approach viable without imposing an undue burden on the operations team. Otherwise the pace of microservices development could overwhelm the ops team with deluges of exceptions or incidents that have to be dealt with.
Istio has also been designed to help microservice developers with features to support productivity and incorporate resilience against network failures. There are tools supporting A/B testing, which can work particularly well for the small components of microservices because it allows alternative versions to be compared readily. It also supports testing via rapid fault injection.
Google’s Apigee, the API management company it acquired in September 2006 for $625 million, is also playing an important role in the hybrid cloud initiative as the adhesive that will connect legacy workloads to the more modern applications that will come on stream.
There is little doubt there is plenty of commitment and a strong all-round tool set behind the combined hybrid cloud. There is still the question whether it is too little, too late given the formidable challenge it faces, with some work to do convincing Cisco’s enterprise customers to come on board. They have after all been diverted towards AWS and Azure over the past year after Cisco’s presumed withdrawal as a direct cloud player and will want to be sure that this new hybrid cloud really is going to fly.