The dramatic changes that are currently sweeping through the storage industry are once again preparing to reshape the look of tomorrow’s enterprise data centers. Among these changes, features and/or products like high availability, solid state drives (SSDs), server virtualization and thin provisioning are emerging as the predominant ones that IT managers are well under ways towards wide spread adoption. But as the implementation of these features begin, they create new storage management ‘gotchas’ that recent enhancements to Symantec’s Veritas Storage Foundation, Veritas Cluster File System and Veritas Cluster Server are designed to help address.
Every IT manager likes the benefits that any one of these four new features and/or products provides. High availability removes ongoing concerns about unexpected application downtime and rapid recovery. SSDs crank up the throughput on performance intensive applications to unprecedented levels. Server virtualization reduces server hardware footprints, decreases energy consumption and improves hardware resource utilization. Thin provisioning minimizes over allocation on storage servers.
However each of these respective technologies creates equally vexing storage management “gotchas”. For example:
- To maintain high availability, both the application and the application’s data must fail over to an alternate server and/or storage system in the event the production server fails. However doing so in a cost-effective, and timely, manner without creating inordinate complexity is easier said than done.
- Purchasing SSDs is one thing; knowing which LUNs consist of SSDs and then placing the appropriate application data on those LUNs is an entirely separate challenge.
- Thin provisioning on storage systems initially works great for new applications. But as data is stored and then deleted on these volumes, there is no way to recapture that storage capacity.
- Server virtualization operating systems tend to offer only rudimentary storage management features and one of their most notable failings is their inability to provide non-disruptive storage administration and manage storage for the individual virtual machines instead of treating all virtual machines equally.
So in response to the storage management “gotchas” that are associated with these new features and products, Symantec yesterday announced a number of enhancements to its existing product lines that should help mitigate these concerns.
First, in regards to high availability, Symantec is removing a number of the middle steps that were previously required to failover a single instance of Oracle, DB2 or Sybase from one server to another. Currently to failover a database instance from a failed server to an active server within a clustered configuration that uses Veritas Cluster Server and Veritas Storage Foundation, the following automated steps occur:
- Detect the production server has failed
- Unmount the file system on the failed server
- Deport the disk group(s) on the failed server
- Import the disk group(s)
- Mount the file system
- Start the application
- Enable the clients to re-connect to the application
Symantec’s integrated storage management with its Veritas Cluster File System and Veritas Cluster Server opens the door for organizations to do sub-minute application failovers using eliminating the need to perform Steps 2 – 5 (above).
The Cluster File System enhancements may be one that organizations find most appealing. Using this feature, organizations can deliver concurrent access to storage for up to 32 nodes while enabling any individual production node to failover to a standby node in less than a minute (the time it takes to restart the application) while keeping the complexity (including database management) associated with this configuration to a minimum.
To account for the new challenges that SSDs present, Symantec updated its Veritas Storage Foundation so it can uniquely recognize which LUNs presented by the storage providers use SSD beneath the covers. To accomplish this, many storage providers wrap metadata around individual LUNs to distinguish one type of LUN from another. These storage providers now include specific metadata that can inform a volume manager as to what type of disk or features that the LUN presented to it possesses. With these latest enhancements, Storage Foundation can now recognize four (4) types of LUNs: SSD, thick (or fat), thin, and thin-reclaimable.
This serves two purposes. First, Storage Foundation can now automatically recognize what types of LUNs are presented to it and put them in the appropriate storage pool. So administrators can configure Storage Foundation to automatically recognize a LUN presented to it as an SSD LUN and put it in a Tier 0 storage pool. Second, when Storage Foundation is used in conjunction with Veritas File System and its Dynamic Storage Tiering (DST) feature, administrators can create policies that automate the placement of files on the appropriate tier of disk according to what tier of disk that file requires.
Closely tied into this automated LUN recognition enhancement are new capabilities in Veritas Volume Replicator (VVR). Storage Foundation has for some time offered its Veritas Thin Reclamation API that allows Storage Foundation to reclaim freed storage capacity on storage systems with thinly provisioned LUNs while SmartMove has been available for local replication and migration. Both of these features are now extended to VVR so it can synchronously or asynchronously replicate data over any distance but only move blocks that contain data whether they reside in thin or thick volumes.
Now organizations can use either fat or thin volumes at the source and either replicate or migrate the data on these volumes to a target with thinly provisioned LUNs and only consume the amount of storage capacity that is equal to the actual amount of data on the source volume. This represents an improvement from the past when VVR essentially did a block by block move of data from one volume to another and was essentially oblivious to whether or not data actually resided on a specific block and gives VVR a decided advantage over competing block-based replication solutions in this area.
Using this new technique, not only will organizations consume less storage on the target side, but the initial copy of the volume should alo occur more quickly. In this first release, VVR will work with any storage that supports thick or thin volumes. However, the Thin Reclamation feature will currently only work with thinly provisioned volumes on 3PAR, HDS, and IBM storage servers since it requires specific integration with their controllers. However certifications with every other major storage provider are in the works and will be supported shortly.
The final item worthy of mention is that Symantec is now officially supporting the movement of Storage Foundation into the Microsoft Windows 2008 Hyper-V parent partition. I wrote about this enhancement in a series of blogs this past summer so it is not “news” per se since Symantec was previously certified with Microsoft Windows Server 2008 and has supported it for some time. But now with this latest release, the Hyper-V parent can take full advantage of features such as thin provisioning VVR and thin reclamation. In so do
t provides administrators new storage management options for Hyper-V child partitions including:
- Advanced, centralized storage management in the Hyper-V parent partition instead of having to manage storage within each child partition
- Dynamically growing and shrinking child partition volumes online
- Provide snapshot capabilities for entire child partitions (system/boot and application data)
- Extends Veritas Dynamic Multi-pathing (DMP) to individual child partitions so each one child partition has more control over path failover and load balancing to specific network paths to storage.
The rapid rate at which change is occurring in the storage industry is creating a plethora of new options for organizations to lower their storage costs while ratcheting up application performance. Yet in order to effectively manage new features and/or products like high availability, server virtualization, SSDs and thin provisioning on a large scale, enterprises need the appropriate management software to do so. These latest enhancements in Veritas Cluster File System, Cluster Server and Storage Foundation should help organizations do exactly that.