By Carryl Roy published: Wednesday, February 03 2010
VSM: Isilon is a publically held company based in Seattle. Can you give us an overview of what your company does?
PR: Simply put, Isilon makes file-based data storage systems that enable the user to “pay as you grow,” adding performance and capacity on-demand as needs change over time. In analyst terms, Gartner refers to us as “the market leader in scale-out NAS.” This means that in contrast to traditional SAN and NAS, which require a user to add more and more disks – or scale up - behind a single “head” or server, eventually creating a performance bottleneck or a capacity imbalance, our systems scale out, with each node linearly increasing performance and capacity to create a much more flexible, cost-effective solution. Isilon’s “secret sauce” is our OneFS operating system, which we combine with the latest in industry-standard hardware – for instance, Hitachi 2TB disk drives and Cisco’s InfiniBand interconnect for intracluster communication – to create a complete storage system. From virtualization to nearline archive, media and entertainment to life sciences and on to manufacturing, we deliver our systems worldwide to help our customers focus on their business, not on managing their storage.
VSM: What benefits does scale-out NAS provide to enterprise data centers?
PR: Well, essentially scale-out NAS enables businesses to greatly increase data access across their organizations so they can extract greater value from mission-critical information, in less time, and accelerate business growth. Because scale-out NAS is very easy to manage and enables businesses to buy only what they need when they need it – then seamlessly add performance and/or capacity as needs change – it significantly reduces capital and operational costs. In short, it’s faster, more scalable, easier to use and more cost-effective than traditional SAN or NAS systems.
VSM: What role does physical storage play in virtual environments?
PR: Physical storage plays the same role in virtual environments as it does for traditional servers: as the repository of application data. The only real difference – and it’s a key difference – is that there are often many more “servers,” with a much greater diversity of workloads among them, attached to the storage in a virtual environment than you’d typically see in a non-virtualized infrastructure. Users often experience the effects of this for themselves when they first move to virtual servers, when they find that their disk usage patterns, performance metrics and oftentimes their overhead for backups and other automated process change dramatically.
VSM: What are some of the roadblocks enterprises experience deploying traditional storage in virtual environments?
PR: Well, many say that virtualization solves IT’s challenges, and that from now on, IT will be relegated to a minor cost center at each organization. Okay, that sounds great. But Gartner is reporting that only approximately 15 percent of workloads are actually virtualized, so there must be a bottleneck, right? What is holding back widespread adoption? It’s traditional storage. As I mentioned before, virtualization changes how storage gets used, and with older SAN and NAS architectures the sought-after benefits of server virtualization are inherently limited. Traditional storage requires over-provisioning, adds complexity, and requires time-intensive management processes when paired with virtual servers. Storage and IT admins are left with a complicated mish-mash of storage volumes in virtualized environments, and that complexity grows as virtualized workloads grow, requiring additional costs for hardware and the staff that administer it. Basically, organizations will never realize the potential of server virtualization with traditional storage. It will always bottleneck at either the performance or capacity axis and regardless of the bottleneck, its complexity and cost cancel out the very benefits virtualization aims to achieve.
VSM:Why is scale-out NAS a better alternative?
PR: Because scale-out NAS eliminates the challenges of scaling virtualized workloads. Let me go into some detail here. With traditional storage, admins are forced to make a choice – over-provision the storage or lose either performance or utilization. That’s a no-win scenario. On the other hand, scale-out NAS automatically uses all available storage, processing, and network resources that are part of the system, so these resources scale transparently and on-the-fly. This unique capability enables an IT admin to provision now for current workloads and scale the entire environment as needed over time without over-provisioning. As the system scales, the IT admin still enjoys the convenience of a single file system – a single mount point/share for users and applications and a single point of administration – regardless of system size.
With scale-out NAS, it’s no longer necessary for a storage admin to configure multiple LUNs or volumes for a set number of VMs. There is no need to group VMs by potential and peak throughput and I/O requirements. It’s no longer necessary for the storage admin to constantly move VMs between LUNs and volumes, attempting to adjust for changing performance requirements while also trying to maximize utilization of the underlying storage. With scale-out NAS, an admin can deploy a diverse set of workloads and allow the storage system to automatically adapt to those workloads without any manual intervention – just as that same admin allows VMs to be automatically balanced.
In short, the inherent benefits of a scale-out architecture mirror the desired benefits of server virtualization, allowing users to get the most from both their storage and their investment in virtualization.
I remember one fellow who came to our booth at least five times. The first time he was skeptical, but curious. By the end of the show he was hooked, staying to the final bell as he checked out the many ways his application could be designed differently without the limitations of traditional storage.
