Windows 8 Server: Interface Core and New File System
Two new information from the time of Windows 8 Server are interesting, a consolidation and change in interface and another for being a “news” which was already expected from the Longhorn (Windows 2008 RTM).
This change is significant, though not new to already be present in Windows 2008 R2. But now the Server Core will be the default interface instead of the full interface on Windows servers 8.
With this change we see as being well reasoned and accepted by customers using a Windows OS with less consumption of memory (512 Kb in Core against 1.5 GB with the graphical interface).
But in addition has been added the possibility to change between the GUI-mode Setup and the Core, which today is not possible on Windows 2008 R2. This will, including is highlighted in the ad, which an administrator can install the graphical interface to configure the server and after finally returning to the Core, which will be very good for those who not know PowerShell.
New filesystem ReFS
Since the Longhorn that spoke of a file system based on database and that has been tested and had the code name WinFS. Put the database model is not as the files for various reasons, but mostly in the form of storing data that differs in many documents.
I often hear people saying that SharePoint saves files in the database, but are data in structured and unstructured binary formats not as is the case of a disk. For example, a doc/gif/jpeg/pdf has start and end in file clustering with application-defined content, while on a file system as NTFS encryption Data Guard for each file and unstructured data such as the Shadow Copy (VSS).
What the ReFS will aggregate database concepts is not BLOB or CLOB format storage, but structuring the file system.
This is easy to understand when we think that in the original NTFS guarding the file metadata in a “table” where there was date created, name and other common data, which will be linked lists of ACL permissions/ACE with the “ID” of the file. With the passage of time and developments emerged many other tables such as encryption, compression, BitLocker, Shadow Copy, etc. With this NTFS eventually dispelling in a series of “tables” that deal with particular metadata. Imagine a disk where NTFS is VSS enabled with encryption and how many different information will be spread among several tables specific to each resource.
Already in ReFS will be used the primary key contents to a file and only one ID and tables will be unified with the concept of “Key Value” common in applications that use the array as.NET and Java.
Below you can see an example that the product team released where the allocation table contains only the ID and reference of physical blocks on disk, a table contains the basic metadata and file another all “Key Values” together.
The concept of “Key Value” is very useful because we can represent any additional information without having to create separate tables. See the example below, the theoretical course of how to represent the improvement.
Note that NTFS model we have a “table” that represents only encryption and for each recovery agent repeats the data. Multiply that for each type of information that a file stores in NTFS.
In model based on structures of tables all information are in a single place based on the “Key” and the value holds the details of that information, reducing the number of tables to control.
This will reduce the fault surface because they are not x tables (or blocks) to save and retrieve the data, being simpler to SO join the information and keep your backups up-to-date (replicas).
Note: Not possible converter file system, it´s necessary move and reformat.