Q&A: Enterprise Storage Challenges
More data means more storage -- and more headaches. We look at today’s storage challenges, the impact of open source storage solutions and server virtualization, and the storage trends to watch this year.
More data means more storage -- and more headaches. From managing storage sprawl to finding open source storage solutions that scale, data center managers have their hands full. We examine today’s storage challenges, the impact of open source storage solutions and server virtualization, and the storage trends to watch this year with Jack O’Brien, senior director of marketing at Gluster where he leads all marketing and product management activities.
Enterprise Systems: What are the biggest challenges enterprises are facing in their storage environment?
Jack O’Brien: There are a host of challenges, but essentially all paths lead back to the exponential growth of digital content. That growth first poses scalability challenges, which go hand-in-hand with performance -- even for applications that are not considered high-performance computing (HPC). Scale then leads to management challenges driven by storage sprawl and a lack of tools to simplify the task. With capacity growing unabated, it’s a constant struggle to control costs. Disk drives ride a steady slope of higher capacity and lower cost, yet the systems and software do not enjoy the same economics and vendors continue to lock customers in.
The other trend causing headaches for storage administrators is the acceleration of server virtualization adoption. Virtual machine environments are dynamic and deployments of hundreds -- or thousands -- of virtual servers puts a serious strain on the infrastructure in terms of performance, capacity, and management. Virtualization is also directly linked to the emergence of cloud computing. Beyond the hype of cloud computing are real problems that data center managers and architects are working to solve, namely predictable service levels, multi-tenancy, and security.
Is the explosion of unstructured data a concern for data center managers, and if so, how are they trying to deal with this issue?
Unstructured data is estimated to account for 80 percent of storage capacity and growing at over 50 percent per year, so managing this explosion is definitely a front-burner issue for data center managers. Traditional storage solutions were not designed for these new challenges and data center managers are looking for alternatives. In the past, the response was to “throw hardware at the problem” which led to storage sprawl and complex management issues.
Data center managers are turning to systems that scale to hundreds of terabytes, or even multiple petabytes, within a global namespace. Technologies that reduce the amount of storage capacity required, such as deduplication, compression, and thin provisioning, are maturing rapidly and becoming widely adopted. Data center managers are working smarter instead of harder.
What are the benefits and drawbacks for organizations adopting open source storage solutions?
The combination of open source software and commodity hardware has transformed multiple segments in the IT industry; Linux for servers, MySQL for databases, JBOSS for Web services, and many other examples. Customers adopt these solutions because they solve problems, lower cost, and reduce lock in.
Open source technologies have entered the mainstream in the enterprise and the industry is well past the point where open source was looked upon with skepticism. In fact, we often see customers with a mandate to adopt open source to accelerate these benefits. There aren’t any inherent drawbacks to a solution just because it leverages open source; it’s a matter of finding a solution to a problem.
How is the storage industry adapting to the increasing adoption of server virtualization technologies?
Scrambling to catch up, in most cases, because server virtualization has a head start. Storage vendors are all actively developing optimized features and increasing integration with server virtualization solutions. Likewise, server virtualization vendors are investing in capabilities to better manage and integrate with storage solutions. Eventually, the industry will gravitate to a model in which a truly virtualized storage infrastructure complements the virtual server infrastructure.
What are the benefits of storage virtualization, and how are storage administrators warming to this technology?
Eventually we will get to the point where the benefits that have been realized on the server side -- improved utilization, flexibility, and cost reduction -- will be established on the storage side. Server virtualization has established a track record and storage administrators are eager to adopt solutions that capitalize on the same principles. Anything that simplifies the management of large pools of storage, improves utilization, and lowers cost is welcomed. This is why we will ultimately see both storage and server infrastructure virtualized and efficiently managed.
How quickly are enterprise customers adopting clustered, or scale-out, storage solutions?
A: Enterprise customers are quickly turning to the scale-out model for storage, and it’s being driven by several factors. History is again a guide here: everyone in IT is familiar with how scale-out computing has grown to a dominant presence in the data center (with open source software playing a big role). At one point, the model was confined to HPC applications, but today it is firmly established in the commercial enterprise and is poised to take off for storage. Traditional network attached storage (NAS) customers are pushing against scalability and performance barriers -- and turning to clustered storage to break through. Managed hosting providers are applying their expertise in clustered computing to deploy clustered storage. Cloud storage solutions, whether they be from service providers or deployed internally, are naturally drawn to a clustered pool of storage resources that can be centrally managed.
Of course, the change won’t happen overnight, and there are many use cases in which traditional NAS and SAN architectures will be employed (just as mainframes have not disappeared). But even customers who purchased a SAN because it was perceived as the only way to get the reliability and performance they required are looking at the landscape and seeing if they can achieve the availability, performance, and manageability they need at a fraction of the cost.
What do you think will be the most important trend to watch in the storage industry in 2010?
I think the most important trend to follow is how the storage industry addresses the gap between server virtualization and storage virtualization. Tactical and strategic moves will both be important and we will see lots of activity at both levels. The established vendors already have company with many emerging startups. I think storage virtualization will be the number one topic in the overall virtualization discussion.
Given the competition in the storage market, what problems does Gluster solve, and how is your solution unique?
Gluster is exclusively focused on simplifying the task of storing and managing the explosion of unstructured data while transforming the economics of storage. The Gluster Storage Platform has a unique architecture that eliminates the need for a metadata server and employs a modular file system design that runs in userspace. This means we can scale linearly from a few terabytes to hundreds of petabytes without the bottleneck issues or reliability concerns of a metadata server. Gluster Storage Platform can be configured for excellent performance across a wide range of workloads. The Gluster Storage Platform is easy to install and manage. Creating a clustered storage volume is as simple as plugging in a USB stick and going through a two step installation process. The economics are extremely compelling. Gluster subscriptions plus commodity hardware offer high value at a lower cost than proprietary solutions.