Читать книгу VMware Software-Defined Storage - Martin Hosken - Страница 9

Where the data lives! That is the description used by the marketing department of a large financial services organization that I worked at several years ago. The marketing team regularly used this term in an endearing way when trying to describe the business-critical storage systems that maintained customer data, its availability, performance level, and compliance status.

Since then, we have seen a monumental shift in the technologies available to vSphere for virtual machine and application storage, with more and more storage vendors trying to catch up, and for some, steam ahead. The way modern data centers operate to store data has been changing, and this is set to continue over the coming years with the continuing shift toward the next-generation data center, and what is commonly described as software-defined storage.

VMware has undoubtedly brought about massive change to enterprise IT organizations and service-provider data centers across the world, and has also significantly improved the operational management and fundamental economics of running IT infrastructure. However, as application workloads have become more demanding, storage devices have failed to keep up with IT organizations’ requirements for far more flexibility from their storage solutions, with greater scalability, performance, and availability. These design challenges have become an everyday conversation for operational teams and IT managers.

The primary challenge is that many of the most common storage systems we see in data centers all over the world are based on outdated technology, are complex to manage, and are highly proprietary. This ties organizations into long-term support deals with hardware vendors.

This approach is not how the biggest cloud providers have become so successful at scaling their storage operations. The likes of Amazon, Microsoft, and Google have scaled their cloud storage platforms by trading their traditional storage systems for low-cost commodity hardware, and employed the use of powerful software around it to achieve their goals, such as availability, data protection, operational simplification, and performance. With this approach, and through the economies of scale, these large public cloud providers have achieved their supremacy at a significantly lower cost than deploying traditional monolithic centralized storage systems. This methodology, known as web-scale, is addressed further in Chapter 6, “Designing for Web-Scale Virtual SAN Platforms (10,000 VMS+).”

The aim of this book is to help you understand the new vSphere storage options, and how VMware is addressing these data-center challenges through its software-defined storage offerings, Virtual SAN and Virtual Volumes. The primary aim of these two next-generation storage solutions is to drive efficiency through simple, less complex technologies that do not require large numbers of highly trained storage administrators to maintain. It is these software-defined data-center concepts that are going to completely transform all aspects of vSphere data-center storage, allowing these hypervisor-driven concepts to bind together the compute, networking, and software-defined storage layers.

The goal of software-defined storage is to separate the physical storage hardware from the logic that determines where the data lives, and what storage services are applied to the virtual machines and data during read and write operations.

As a result of VMware’s next-generation storage offerings, a storage layer can be achieved that is more flexible and that can easily be adjusted based on changing application requirements. In addition, the aim is to move away from complex proprietary vendor systems, to a virtual data center made up of a coherent data fabric that provides full visibility of each virtual machine through a single management toolset, the so-called single pane of glass. These features, along with lowered costs, automation, and application-centric services, are the primary drivers for enterprise IT organizations and cloud service providers to begin to rethink their entire storage architectural approach.

The next point to address is what software-defined storage isn’t, as it can sometimes be hard to wade through all the marketing hype typically generated by storage vendors. Just because a hardware vendor sells or bundles management software with their products, doesn’t make it a software-defined solution. Likewise, a data center full of different storage systems from a multitude of vendors, managed by a single common software platform, does not equate to a software-defined storage solution. As each of the underlining storage systems still has its legacy constructs, such as disk pools and LUNs, this is referred to as a federated storage solution and not software-defined. These two approaches are sometimes confused by storage vendors, as understandably, manufacturers always want to use the latest buzzwords in their marketing material.

Despite everything that has been said up until now, software-defined storage isn’t just about software. At some point, you have to consider the underlying disk system that provides the storage capacity and performance. If you go out and purchase a lot of preused 5,400 RPM hard drives from eBay, you can’t then expect solid-state flash-like performance just because you’ve put a smart layer of software on top of it.