Gluster brings clustered NAS storage to VMware

Virtualization has changed the way we look at the data center and the component pieces that comprise it. With a massive consolidation effort taking place, servers with large processing power and memory have become more of the standard in a virtualized data center. Multiple processors in a single server will no longer cut it; now we require multicore, multisocket processors to properly push these virtualization and cloud infrastructures. Memory size has also tried to keep up pace, growing from an average 2GB to 4GB of RAM per server to between 64GB and 128GB.

But as the momentum to virtualize the data center continues, storage demands are also increasing. Virtualization and cloud computing are increasing the demand for storage capacity as the amount of data continues to explode and grow exponentially. And the cost of providing and managing SAN storage isn’t getting any cheaper.

[ Also on InfoWorld: Headed to VMworld? Check out the changes happening with VMworld 2010 | Make sure to keep up with all the latest virtualization news with InfoWorld’s virtualization newsletter. ]

To address the growing concerns around storage, this week, Gluster introduced VMStor, a NAS solution for virtual machine storage. The company said it is looking to NAS technology in order to offer a cheaper alternative to SAN when trying to scale out a cloud or virtualized data center. Gluster VMStor is said to address that challenge by scaling to multiple petabytes and storing an unlimited number of VM images. It also offers capabilities around VM-level snapshots and single mouse-click backups.

So how is virtualization and cloud computing impacting storage? And what should organizations be asking right now as they confront virtual machine storage issues? And how does NAS compete with SAN in this virtualization and private/public cloud environment? To get answers to these and other questions, this week, I spoke with AB Periasamy, CTO of Gluster.

InfoWorld: How is the introduction of virtualization and virtual machines impacting storage and the IT environment?

Gluster: Innovation for server virtualization has outpaced corresponding technology on the storage side, thus storage administrators are playing catch-up. Traditional silo-ed storage architectures lack the scalability and flexibility required for a modernized, virtualized data center. Adding virtual machines to an environment requires additional storage capacity to alleviate the performance degradation that arises with a virtual server environment.

The two biggest challenges are management and performance. An environment with a thousand virtual machines attached to a SAN needs to have a thousand LUNs provisioned and managed. In a typical environment, an administrator would provision storage and I/O every time a VM is created, requiring complex management. This level of management complexity puts a huge strain on the infrastructure. With many virtual machines trying to access the same data, bottlenecks arise that can significantly impact application performance. Dealing with these issues keeps storage administrators up at night.

InfoWorld: Talk about some of the IT trends that you have seen that have helped push the adoption of virtualization technologies.

Gluster: Data center managers recognize that the utility model for computing is the most efficient way to deliver services. Virtualization is the key ingredient for flexibility and automation. The promise of decreased costs and increased manageability is the driving force behind any enterprise adopting virtualization/cloud computing technologies. Cloud computing is the primary driver and is an architectural evolution of the data center that when tightly coupled with the need for virtualization is now the focus of innovation.

InfoWorld: And how are enterprises handling things today?

Gluster: Existing infrastructure is strained, particularly on the storage side as noted as that element has lagged behind, but many new technologies are emerging. Several trends, such as the explosion of unstructured data and the emergence of cloud computing, have shined a spotlight on the gap and woken many to the realization that it is holding the industry back from achieving a fully virtualized data center. The virtualization vendors are doing their part to rapidly adopt themselves to the cloud requirements with improved management tools and new APIs.

Many enterprises are moving to public cloud platforms such as those provided by Amazon Web Services and Rackspace. New players such as Eucalyptus and Cloud.com are introducing cloud management software stacks. Storage vendors, both existing and new, are making major investments to close the server-storage virtualization gap. Storage was often considered an afterthought and systems designed for transaction-oriented databases were a poor match for new demands.

The storage layer has to scale linearly in capacity and performance; throwing hardware at the problem is not a solution. Storage virtualization demands an entirely new software-based approach that leverages a scale-out architecture to deliver petabytes of capacity and gigabytes per second of throughput.

InfoWorld: What are some of the questions enterprises should be asking while they are trying to figure out how to address their virtual machine storage?

Gluster: The key to a utility or cloud environment is how effectively resources can be pooled and shared, so this is the first criteria enterprises should address. Scalability is critical as well — not just the ability to store large amounts of data, but the ability to respond to dynamically changing workloads. Also, manageability is extremely important; enterprises need to look for simple management that automates operations as much as possible. Finally, the economics need to be addressed; commodity storage solutions are now enterprise ready and should be considered for enterprise deployment. There is growing evidence that cloud infrastructure will rely heavily on commodity hardware and open source software.

InfoWorld: With your latest VMStor offering, you seem to be betting on what you call scalable NAS. Where do you see NAS fitting into the virtualization and cloud landscape?

Gluster: We think the future of virtual and cloud storage is NAS. NAS scales to petabytes, enables shared data access across many clients, and is easy to use, cost effective, and high performing — everything that is required to effectively manage data in VMs.

Unstructured data is exploding and it’s predicted that by the end of 2010, 1,200 exabytes of data will be created. Scale-out NAS solutions can scale seamlessly and pool disk and memory resources under a global namespace virtualizing the underlying hardware whether you require a thousand virtual disk images, a petabyte of data, or both. This provides an ideal virtual storage environment to complement virtual server deployments. Additionally, NAS volumes can be mounted simultaneously across thousands of servers, providing both the hypervisor and cloud applications to share the same storage, allowing VM migration across a large pool of servers without concerns for storage access.

InfoWorld: For years now, organizations with virtualized environments have been told to leverage the power of SAN, but you guys are talking NAS vs. SAN. Why is that?

Gluster: SAN solutions have some inherent limitations for scalability, manageability, and ease of sharing data. They can also be expensive, even with the availability of iSCSI-based solutions, and require fairly sophisticated administrative expertise. A cloud environment where hundreds of VMs can be provisioned in minutes is extremely dynamic. SAN administration requires significant expertise such as the need to use multiple extents to create a VMFS volume larger than 2TB or implementing raw device mappings (RDM) for performance.

Alternatively, NAS solutions are easily shared and a single mount point can be addressed by hundreds or thousands of clients. VM disk images are just files and NAS is optimized for large-scale storage of files making it an ideal match for virtual machine storage. Key features such as snapshots, thin provisioning, and backup are all supported, and previous questions about performance are no longer relevant with NAS supporting 10 Gigabit Ethernet and InfiniBand.

InfoWorld: At what point do you see cloud storage becoming mainstream?

Gluster: We think that it’s interesting that even though the hype surrounding cloud computing is probably at an all-time high, there are a lot of use cases that we are seeing. We definitely see both private and public cloud storage in production. The industry knows that mainstream adoption is inevitable, real proof points are supporting the momentum and this is a case where the industry will figure things out quickly.

Special thanks to AB Periasamy, CTO of Gluster, for speaking to me about the topic of open source NAS storage in a virtualized and cloud enabled data center.

This story, “Gluster brings clustered NAS storage to VMware,” was originally published at InfoWorld.com. Follow the latest developments in virtualization and cloud computing at InfoWorld.com.