Archive: Posts Tagged ‘Microsoft’

Thoughts on Server 2012 R2 Scale-out File Server and Provisioning Services

1 comment September 17th, 2013

When I read about all the new features and enhancements that comes with the Windows Server 2012 R2 release, I started to realize how much of it could be used to optimize and save costs when used in a PVS platform. Ill try to go through a few of the enhancements and features and share my thoughts of how they could be used.

Failover Clustering and Scale-out File Server

Ive never been a fan of Microsoft Clustering. The experiences i have and what ive heard from others using it is thats its very cumbersome to manage and troubleshoot. When something goes wrong its a pain to get it working again.
Well with failver clustering and Scale-out File server in server 2012 this seems to have been improved.

Now you have features like Cluster-Aware Updating that enables software updates to be applied automatically to the operating system or other system components on the nodes of a failover cluster while maintaining availability during the update process. This will reduce maintenance time by automating what was previously a very repetitive task.

With the introduction om SMB 3.0 and Storage Spaces you get performance, scalability and functionality that used to only be available in big expensive SAN solutions.

The Scale-out File Server (SoFS) role uses SMB 3.0 to enable you to build Active-Active fileclusters. SoFS will automatically redired I/O over SMB to an optimal node in the cluster. When the client connects to a share in the cluster it will connect to any of the nodes. However the file the client wants to open might actually be stored on one of the other nodes in the cluster. SoFS will detect this and automatically transition the SMB connection to the node where the file resides.

With the Server 2012 R2 release Failover Clustering and SoFS also supports data deduplication which can save you lots of storage space when your vDisks are stored on a SoFS share.

Storage Spaces

Until Windows Server 2012 was released, implementing storage virtualization required purchasing proprietary third-party SAN solutions that were expensive and required using their own set of management tools. Such solutions also required special training to implement, manage, and maintain them effectively. Storage Spaces, first introduced in Windows Server 2012, was designed to make storage virtualization affordable even for small businesses. Storage Spaces is simple to deploy and manage, and it can provide businesses with shared storage that can grow on demand to meet an organization’s changing needs.

 With Storage Spaces in Windows Server 2012 R2 Microsoft enables you to create a tiered storage that transparently delivers an appropriate balance between capacity and performance that can meet the needs of enterprise workloads. The result is that the workload’s most frequently accessed data (the working set) will automatically be stored on the SSD tier while the rest of the workload’s data is stored on the HDD tier.

Data deduplication

Data deduplication was introduced in Windows Server 2012 to help enterprises cope with exponentially increasing growth of data storage in their environments. Data deduplication allows Windows Server 2012 to store more data in less physical space to optimize the capacity of their storage fabric. Data deduplication is highly scalable, resource efficient, and nonintrusive in Windows Server 2012, and can run on multiple volumes simultaneously without affecting other workloads running on the server. Checksums, consistency, and identity validation are used to ensure data integrity, and redundant copies of file system metadata are maintained to ensure data is recoverable in the event of corruption.

Sadly this wont help save storage in you PVS environment as you might think. In server 2012 data deduplication can only be used with files that are closed. Your PVS vDisks are likely to be open most of the time as the target devices is running from them.

In server 2012 R2 thou Microsoft added a few improvements to the way data deduplication works. When used in a Failover Cluster data deduplication will work on open files aswell!

SMB 3.0

SMB 3.0 is at the core of the new SoFS functionality introduced previously in Windows Server 2012 and enables file-based storage solutions (file servers) to provide network storage for servers that have similar performance to expensive, proprietary SAN solutions. Using SMB 3.0 and the Windows Server 2012 platform, organizations can build low-cost scale-out storage fabrics that can meet the needs of a broad range of private and hosted cloud solutions.

Some of the key features of SMB 3.0 include:

  • SMB Scale Out
    Allows you to create file shares using Cluster Shared Volumes (CSV) to provide simultaneous access to data files, with direct I/O, through all nodes in a file server cluster. This results in improved utilization of network bandwidth, load balancing of SMB 3.0 clients, and optimized performance for server applications.
  • SMB Transparent Failover
    Allows you to perform hardware or software maintenance of nodes in a clustered SMB 3.0 file server (an SoFS) without interruption to server applications storing data on the file shares. If a hardware or software failure occurs on a cluster node, SMB 3.0 clients can transparently reconnect to another cluster node without interruption as well.
  • SMB Direct
    Supports using network adapters that have Remote Direct Memory Access (RDMA) capability, which allows them to function at full speed with very low latency and very low CPU utilization. When used with workloads, such as Hyper-V or SQL Server, the result is that a remote SMB 3.0 file server can resemble local storage in its performance.
  • SMB Multichannel
    Provides aggregation of network bandwidth and network fault tolerance when multiple paths are available between the SMB 3.0 client and the SMB 3.0 file server. This results in server applications taking full advantage of all available network bandwidth and being more resilient to network failure.

All these features boosts the SMB protocol performance significantly when used with supported hardware and adds resiliancy with multichannel and transparent failover. This for me takes the SMB protocol into the cloud era and enables it to be used in more critical infrastructural functions while still being easy to mange and setup.

If you put all this together and build a SoFS cluster to store your vDisks you get a centralized vDisk store and dont need to sync your vDisks between servers using local storage. You will save even more storage with deduplication as the vDisks is likely to be very similar to eachother.

Windows Server 2012 R2 SoFS is an inexpensive, valid and fully featured storage solution that can compete with many enterprise storage solutions out there.

There are many more features and enhancements in the new R2 release i havnt talked about as i think these are the most important ones that relate to Provisioning Services.