The continued growth in the amount of data (structured and unstructured) and the need to analyze it real time has become a must need core competency for leading Enterprises. This market trend has accelerated the adoption of the Cloud by those Enterprises since it enables them to exploit the agility, flexibility, and simplicity that the Cloud provides as well as focus their attention on the data itself. This enables the Enterprises to achieve their goals economically.
Taking advantage of those Enterprises’ needs, the Cloud Service Provides (CSP) have been encouraging IT managers to move 100 percent of their infrastructure to the C
loud; providing incentives in the form of attractive services and pricing. A good example of this is a service for Enterprises that have a large amount of storage which can’t be sent to the CSP over the Internet. In this case, the CSP sends a truck and takes the physical hard disks ̶ and all for FREE! However, the main question that IT managers need to ask themselves is: “How much it will cost me to get the storage back?” (assuming of course, that the CSP even provides such a service).
The storage scenario, of course, is just one example of the risks that Enterprises can face if they decide not to have any on premise data center and to move all of their IT services to the Cloud. There are, of course, other legal and privacy issues that different companies may have to navigate, and this is why Hybrid Cloud approach provides the most appealing option because it enables complete interoperability and seamless operation between the Enterprises’ private Clouds to its CSP. Not having this is like driving down a one way street without the ability to turn around (and thus, the giving the CSPs more room to increase pricing).
Figure 1: Bringing Together the Best of All Worlds (Source: Gartner)
Although Hybrid Cloud architecture is not new, and in theory it looks simple, it has actually been very complicated to implement, and up to now, all the solutions that the CSPs have provided have not guaranteed seamless interoperability. There are various reasons that this has not yet happened. Some of them software related, i.e., the usage of the exact OS, hypervisor, the software stack and orchestration, and the other have been on the hardware side, i.e., the usage of the exact servers, storage and networking.
However, this situation is about to change for Azure Cloud, at Inspire’17, where several leading OEMs are announcing their Azure Stack systems. The event is being held July 9 – 13, in Washington D.C. This will enable a true Hybrid Cloud between Enterprises’ on premise private Clouds to Microsoft’s Azure Cloud. These systems are based on Microsoft’s “recipe” which guarantees the “must-to-have” seamless interoperability. Moreover, it includes a specific version of Windows Server that is a subset of Windows Server 2016 which supports Storage Spaces Direct and requires a RDMA-enabled networking. A good description can be found at: Enabling Higher Azure Stack Efficiency – Networking Matters.
Figure 2: Lenovo ThinkAgile SXM4200 25U (left) and SXM4200 42U/SXM6200 (right)
Inspire’17 visitors can see Azure Stack Systems at work in Lenovo booth #1214 where they are demonstrating Think Agile SX for Microsoft Azure Stack. This Cloud solution was announced in June’17. It was co-engineered with Microsoft and Mellanox and brings the power and convenience of Microsoft’s Azure Stack using Lenovo’s advanced servers, rated number one in customer satisfaction and reliability. These servers are equipped with the Mellanox’s 10/25GbE RDMA-enabled networking solution with Lenovo’s Networking switches, which offloads network functions from running on the CPU to the IO subsystem. All of which results in much needed higher Cloud efficiency.