Although there are a lot of new
features and functions added in to Windows Server in the past couple
revisions (2012 and 2008 R2) that make the headlines, one of the first
places I like to start is around the things that don’t make the
headlines that are really core technology improvements from which many
of the features are built from. These are technologies that make the
new operating system faster, more reliable, and do more things—but they
aren’t features that you have to install or configure.
Self-Healing NTFS
One of the embedded technologies in Windows
Server 2008 and extended into Windows Server 2012 is self-healing New
Technology File System (NTFS). Effectively, the operating system has a
worker thread that runs in the background, which makes corrections to
the file system when NTFS detects a corrupt file or directory. In the
past when there was a file system problem, you typically had to reboot
the server for chkdsk to run and clean up file and directory corrupt
errors.
This self-healing function
is not something you will ever see running; however, it is an added
capability under the hood in Windows Server 2012 that keeps the
operating system running reliably and with fewer system problems.
Server Message Block 3.0
Introduced in Windows Server 2012 is Server
Message Block 3.0, more commonly called SMB3 or SMB for short. SMB is a
protocol that handles the transfer of files between systems.
Effectively, SMB compresses file communications and, through a larger
communications buffer, is able to reduce the number of round trips
needed when transmitting data between systems.
It is analogous to the difference between the copy command and the xcopy command in DOS. The copy command reads, writes, reads, writes information. The xcopy
command reads, reads, reads information and then writes, writes, writes
the information. Because more information is read into a buffer and
transferred in bulk, the information is transmitted significantly
faster.
Most users on a high-speed local-area network
(LAN) won’t notice the improvements when opening and saving files out
of something like Microsoft Office against a Windows Server 2012
system; however, for users who might be copying up large image files or
data sets between systems will find the information copying 10 to 30
times faster. The performance improvement is very noticeable in
wide-area network (WAN) situations on networks with high latency.
Because a typical transfer of files requires short read and write
segments of data, a file could take minutes to transfer across a WAN
that can transfer in seconds between SMB-connected systems because the
round-trip chatter is drastically reduced.
For SMB3 to work effectively, the systems on
both ends need to be Windows Server 2012, or Windows 8 systems. A
Windows XP client to a Windows Server 2012 server will communicate over
SMB 1.0 for backward compatibility and will not gain from this new
technology. In addition, when talking to Windows 7 or Windows Server
2008 R2 machines they can only negotiate up to SMB 2.1.
Significant to Windows Server 2012 in regards
to SMB is Microsoft’s inclusion of SMB for clustering and replication
technologies built in to the new operating system. It used to be that
if you wanted to setup a clustered server environment, you needed a
storage-area network (SAN) shared stored solution to failover clustered
servers from one to another. With Windows Server 2012, clustering can
be done simply with SMB. Two Hyper-V hosts can be set up, load on the
Failover Clustering feature, set up a basic Windows 2012 file server,
point the servers to an (SMB file share) of the file server, and build
the cluster without a SAN.
Hyper-V
Hyper-V is a technology built in to the core
of the operating system in Windows Server 2008 and expanded in Windows
Server 2012 that greatly enhances the performance and capabilities of
server virtualization in a Windows environment. In the past, virtual
server software sat on top of the network operating system and each
guest session was dependent on many shared components of the operating
system.
Hyper-V provides a very thin layer between
the hardware abstract layer of the system and the operating system that
provides guest sessions in a virtualized environment to communicate
directly with the hardware layer of the system. Without having the host
operating system in the way, guest sessions can perform significantly
faster than in the past, and guest sessions can operate independent of
the host operating system in terms of better reliability from
eliminating host operating system bottlenecks.
As mentioned in the previous section on
Hyper-V and the ability to create a cluster with SMB file storage,
there have been other very significant improvements in Hyper-V,
including the ability to do site-to-site replication of Hyper-V guest
sessions. Instead of purchasing expensive replication technologies,
Hyper-V hosts can be pointed to one another for server to server
replication. A number of other technologies have been greatly improved
in Windows Server 2012.
Storage Spaces
Storage spaces represent a significant
inclusion in Windows Server 2012, effectively the ability to group
together storage space on multiple servers and having them displayed
and accessible as a single storage share. Storage spaces work in the
same way that RAID drive mirroring or RAID striping works, replicating
or striping data across drives for higher availability. With storage
spaces, however, the mirroring and striping can be done across systems.
So, instead of having two drives mirrored in one server (with the
server as a bottleneck or point of failure), the two drives can be in
two separate servers, and the mirroring happens between the two
servers. Or three drives can be split across three different servers
and striped across the three servers providing both storage resilience
as well as server resilience.
The concept of mirroring and striping storage
spaces on small 50GB or 100GB levels doesn’t register as a benefit for
most IT people because mirroring 100GB drives is done all the time
inside a server. But what if you want to create a 10TB video, media,
and large-image storage repository, and that “server” that the 10TB
disks are in fails? What will it take to restore 10TB from tape? Or
what if you want to copy the 10TB to another server? It will take an
extremely long time, which is why organizations buy external SAN
storage, to put large amounts of data and then snapshot that large
amount of data for redundancy.
But what if you create two servers or three
servers or five servers with each 10TB or 5TB and mirror or stripe the
storage across multiple servers? Now you have no single point of
failure, the servers have data mirrored or striped, the servers are
redundant, and you do this all at a fraction of the cost of a SAN or
SAN plus snapshots.
Add in all of this the
flexibility provided by SMB shares as an underlying technology for
high-speed data clustering and site replication, and storage spaces
start to make a lot of sense.
Figure 1. Storage Spaces in Windows Server 2012.
De-Dupe
De-dupe, or data de-duplication is built in
to Windows Server 2012 and is the ability for basic Windows 2012 file
system storage to be de-duplicated to decrease storage capacity
demands. As an example, if a 10TB share of data has the same video 20
times with variations of the video from early draft cuts through near
final and final versions, the video could take up a lot of disk space.
Data de-duplication acknowledges the replication of bits on the disk,
and instead of having the same data multiple times, it flags the data
as duplicates and opens up space for the storage of other information.
In many cases, data de-duplication
has saved 30% to upward of 70%, with averages being somewhere in the
middle. Organizations with 10TB of space are often able to achieve 50%
to 55% space reduction, and no longer have to purchase more storage as
they reach 10TB that might be the limit of their current storage
subsystem. This 50% savings could prevent the organization from buying
more storage for another year or two, potentially a huge benefit to
organizations today.
