Windows Server 2012 supersaver No. 6: SMB 3.0
Microsoft has invested a substantial amount of engineering effort in the modernization of the Server Message Block (SMB) protocol. What started out as a point release to version 2.2 ended up being a major release that deserves the 3.0 label. SMB 3.0 delivers big value in making it possible to use commodity disk drives to deliver high-end storage features. SMB 3.0 shares in Windows Server 2012 support both Hyper-V and SQL Server workloads that previously required block-based storage.
SMB 3.0 includes a number of new components to improve its ability to detect and recover from a lost connection. In the past this relied on a TCP/IP timeout that would typically take up to 20 seconds. SMB transparent failover utilizes a new feature called the Witness service to detect connection failures, then redirect the client to another node. This feature requires clustering in order to fully support the failover and is a part of the Cluster Shared Volumes (CSV) functionality.
Windows Server 2012 and SMB 3.0 include support for Remote Direct Memory Access (RDMA) network adapters. SMB Direct supports most of the latest RDMA adapter types, including iWARP, InfiniBand, and RDMA over Converged Ethernet. SMB Direct works in conjunction with SMB Multichannel to deliver both increased performance and failover capabilities not previously possible over SMB.
SMB Multichannel works behind the scenes to create multiple connections between SMB 3.0 client and server to increase throughput with no user intervention. This feature is enabled by default and has the ability to greatly accelerate everything from simple file copy operations to high-transaction-rate SQL applications. SMB Multichannel does require both systems involved to be running either Windows Server 2012 or Windows 8.
Windows Server 2012 supersaver No. 7: Scale-out file server
One of the problems of building systems to meet a specific workload is what to do when you run out of capacity. This can be particularly bothersome when it comes to a storage system with high availability requirements. Traditionally, you'd have to take the system down, install the new hardware, and reconfigure. Windows Server 2012 addresses this issue on several fronts. Clustering plays a central role here along with thin provisioning of Storage Spaces volumes.
ReFS will support volume sizes up to 2^78 bytes (256 zettabytes) using a 16KB cluster size. Single files can be up to 2^64 - 1 bytes (16 exabytes). The maximum size of a single storage pool is 4 petabytes (PB). Suffice to say, you can create extremely large volumes without the need to have the physical disks to match. As demand for capacity increases, you simply add more disks.
In many cases this can be accomplished without taking a server offline. At a minimum you could take individual nodes in a cluster down without taking the entire cluster offline. The same concept applies to compute power in that you simply add more nodes if the demand for processing power increases.
This story, "7 ways Windows Server 2012 pays for itself," was originally published at InfoWorld.com. Follow the latest developments in Windows and data center at InfoWorld.com. For the latest business technology news, follow InfoWorld.com on Twitter.
Read more about microsoft windows in InfoWorld's Microsoft Windows Channel.
This story, "7 ways Windows Server 2012 pays for itself" was originally published by InfoWorld.