Many enterprises have been watching the development of flash with a high level of interest though they have cautiously deployed it as a storage tier because of its high cost. Symantec Foundation Suite 6.1 takes these concerns head-on by delivering functionality that will inherently change the way enterprises design, implement, and maintain shared storage environments. In particular, Storage Foundation’s new Flexible Storage Sharing (FSS) feature provides organizations the flexibility to non-disruptively put capacity, performance or both in their servers while still making it accessible to all of the applications in that cluster.
Anyone who is at all accustomed to tuning performance in enterprise data centers knows it has been almost impossible to achieve predictable performance in a traditional infrastructure without loading up your Tier 1 storage frame with loads of 15K drives and as many solid state drives (SSDs) as you could afford to buy. The primary motivation for this deployment method was they were following the well-established equation of more spindles = more IOPS. This unfortunately meant many organizations probably added more disk drives for performance reasons as opposed to capacity needs, thus increasing their total cost of ownership (TCO) across the data-center.
To date there really have not been many options for enterprises that wanted low-cost, high-speed, and reliable solutions to deliver high IOPS and low latency without buying loads of SSDs in a multiple form factors available today (All flash memory storage arrays, flash cards in the server, hybrid HDD/flash storage arrays, etc.)
Enter Storage Foundation 6.1 and its new Flexible Storage Sharing (FSS) functionality. This combined with Symantec Cluster File System really offers a unique set of options that not only protect an enterprise’s current investments in servers/storage infrastructure but also enables any future designs to be augmented to reduce overall cost and increase performance.
FSS enables the following functionality
- Global Namespace. Offering a namespace that is currently supported across eight (8) nodes in a cluster, it enables any application in the cluster to quickly access data on any node.
- Local DAS. The flexibility to use internal SSDs, SATA-SSD, SAS and SATA drives inside the server(s) in a shared nothing storage model (i.e. no external SAN required).
- High Performance. Enable the usage of internal, lower cost SSDs for very high performance across the entire cluster while delivering extremely low latency.
- Scalable. If more performance is required, simply add more SSDs into the existing servers. Conversely, if more capacity is needed, simply add more SAS/SATA HDDs. In this way, organizations can get the exact type of storage capacity they need in the amounts they need. Further, since this is a clustered solution, this capacity can be added non-disruptively.
- Protection and High Availability. Clustered CFS provides local, metro and geographically dispersed levels of protection.Application data can have multiple profiles striping across multiple nodes to achieve higher performance, in addition to ability to mirror across multiple nodes to enable high availability locally. Volume Replication functionality can also be enabled in order to provide data replication across geographically dispersed locations.
Implementing FSS gives organizations a high level of investment protection for their existing infrastructure. With the release of Storage Foundation Suite 6.1, Symantec is not saying, “Hey, take our software and buy new servers and new storage to run it on.” Symantec takes a completely different tack by giving organizations the option to “Keep what you have and we can use that too.”
For example, assume you have a three (3) node Linux Cluster attached to shared storage. More often than not these servers are booting directly from this shared storage or off a very small root partition on internal drives on the server.
Most servers probably have 2-8 drive bays free depending on the server model used. Now instead of buying more expensive SSDs for the external storage arrays, organizations have the flexibility to add either SSDs or SAS (15K, 10K or 7.2K) drives to any those server drive bays in the cluster and enable FSS. The file system can deal with the redundancy and performance for those new drives while making their capacity or performance associated with these drives available to any application in the cluster. In so doing the organization just moved its most critical data closer to the application while at the same time probably freeing up some capacity on the shared storage.
FSS in essence creates a mega fast level of tiered storage directly inside each node in the cluster while at the same time requiring an organization to only make minimal or potentially no changes to its existing infrastructure. Using their existing storage or only adding incrementally to it, organizations can transform their “tiered storage infrastructure” into a “hybrid storage environment” so they can easily and cost-effectively tier their data as it ages. This architecture makes it feasible for those organizations with clustered applications that have needed more capacity, more performance or both to have access these features at a much lower upfront cost and in less time than using external, shared storage environments.