Dell has used a couple of software conferences as bully pulpits for some of its own cloud announcements. The company is making a major cloud move, one that they obviously hope will elevate them to the status of a “real” computer company (they rank number three in our surveys as what users think is a “real” player in the space, after IBM and HP, but it’s a distant third). In their effort they’ll partner with both VMware (one conference pulpit) and Salesforce (the other) to offer Dell-branded cloud technology, but they also intend to host open-source cloud offerings (a la Hadoop, perhaps) and even Microsoft Azure.
Dell’s greatest strength has been in the SMB space, and that is also perhaps the best target for cloud services in the near term. Enterprises secure good economies of capital and support scale in their normal data center build-outs, and it’s hard for public cloud services to compete. For the SMB, neither capital nor support economies are easily established, but the latter in particular is problematic because SMBs often can’t attract skilled IT technicians. Remember that Dell also has a professional services arm now, and that means its own support skills likely have a lower marginal cost. All to make their price potentially more attractive.
VMware, meanwhile, is advancing its own cloud position with a Data Director designed to create an enterprise DBaaS model that would also in my view facilitate cloud models where the application or its components ran in the cloud and the data stayed in the enterprise’s own repositories. This would help considerably in building a larger cloud TAM because it dodges the thorny problem of cloud data pricing and security.
In another initiative, VMware has joined with Arista, Broadcom, Cisco, and Emulex to create what they call the “Virtual Extensible LAN) or VXLAN. This is a strategy to add a 24-bit header to a VLAN packet and then encapsulate the whole thing in IP. It would allow the creation of more VLANs with more members and do so using scalable IP rather than Ethernet. VMware will be adding VXLAN support to its Hypervisor and the result would be a more scalable data center and cloud LAN architecture. The four obviously hope this will become a new model for addressing distributed cloud resources.
The initiative is more important for its goal than its methodology. We’re seeing network technology adapting to the cloud. That shouldn’t be surprising, nor should it be happening only now. The network creates the cloud; it’s the binding force that makes not only the resource pool possible but also makes its access possible. The network is the business case, the network is the business. But the network has been silent on the topic of the cloud. Maybe this is a sign that the silence is finally over.
Cisco has also (finally) taken a step to get traction in the service layer, not with Videoscape or another broad-based initiative but in the mobile space. They’ve established a partnership with NEC to sell LTE systems that will include the Cisco/Starent ASR 5000 and the NEC base stations. NEC isn’t a household word in RAN, but that suits Cisco fine; they want to be the kingpin of the deals in the outside-Asia markets anyway. Mobile credentials have been the strongest reason for Alcatel-Lucent’s gain in market share in the router/switching space. Now Cisco hopes to counter the move. The deal puts the most near-term pressure on Juniper, who must now leverage its NSN position better and/or establish the very broad-based service layer strategy that Cisco seems determined to avoid.
That’s a bad choice in my view. NEC doesn’t have the juju in LTE to pull Cisco to the front of the line on mobile deals. What it needs to do is to combine LTE presence with some savvy mobile content monetization. In short, link it to Videoscape. If Cisco can really do that (which it could through the Cisco Conductor XMPP bus and some glue) it would have a truly durable and maybe even compelling mobile position.
Juniper announced an enhancement to its Virtual Gateway (vGW) to provide security for virtualization, and cloud, environments. Juniper has always had a strong security portfolio but it’s only been recently that it’s promoted the tools directly into the cloud data center space. There are security features in the new QFabric architecture that will be fully available late this year, and the new vGW and Junos Pulse strategies both play well with those enhancements and create what is arguably a complete cloud security solution, the most complete on the market. But the elements are coming together slowly, and UBS lowered estimates and its target price for Juniper based in part on macro-economic concerns.
Brocade is also experimenting with a new business model, the “router-as-a-subscription”. The customer gets a router for nothing but pays on a per-port-per-month basis for how it’s configured and used. The model is already being seen by the Street in a bipolar way; some are saying it’s a nervy innovation and others that it’s a sure sign of an industry in its death throes. When you cave that much to cost pressures, the nay-sayers believe, you admit that your pricing power is gone forever.
There’s a decent notion behind this say supporters, even though it’s still cost/defensive in nature. The idea is that by making the router a subscription service you transfer it from the capital budget to the monthly expense budget, which may be attractive from a cash flow perspective (you can write off 100% of expenses but on the average only a quarter or more of capital cost per year). It’s an argument that’s being made in cloud computing, after all. I agree that it’s a clever play, but it’s still an illustration that the enterprise router market is so abysmally price-pressured that you have to play accounting gimmicks to make a sale.
Another interesting development is an announcement by the two big independent DNS players (Google and OpenDNS) that they’ll support geographic (or at least address-hierarchical) extensions to the DNS lookup process to help insure that the users get linked to content caches that are closest to their particular location. This is a pretty significant step for a number of reasons, not the least of which is that these big DNS players are potentially removing a differentiator used by CDN providers. The downside of the idea is that it’s not granular enough to optimize delivery within a metro area, in my view. It’s a way to make “normal” CDN access work better, but not the leading-edge distributed-cache metro-optimizing versions of CDNs. There, operators will have to come up with their own solutions (or rather find vendors who solve the problem for them).
Rounding out the story, Ericsson has introduced some enhancements to its router line, proving out what some of our survey carriers told us about a renewed initiative for Ericsson at the IP layer. The company wants to play on the theme of “premium service” routing but here as with its competitors there’s a fairly limited notion of what a “premium service” really is. For vendors, it’s about transport and connection; for operators it’s about content and mobile/behavioral and (increasingly) the cloud. In our surveys, Ericsson still falls into the noise level for IP-layer competitiveness but there are signs that the company is getting better recognition there.