THE UNIVERSITY OF SCIENCE, VNU-HCM

FACULTY OF ELECTRONICS AND TELECOMMUNICATIONS

DEPARTMENT OF TELECOMMUNICATIONS AND NETWORKS

COURSE
Network Technology

Chapter 1 Local Storage

DATA STORAGE
05
October 18, 2022

Nguyen Viet Ha, Ph.D. Email: nvha@hcmus.edu.vn 2

Storage Devices Storage Devices

❖Combine hard disks and SSDs → low cost

• Operating system, applications, • Not accessed frequently

and frequently accessed data data (e.g., archive
(e.g., databases and shared databases and shared
folders that store user folders that store backups)
documents)

3/63
/50 4/63
/50
 Storage Devices Partitions, Filesystems, and Volumes
Communications interface ❖Storage device is divided into sections
called partitions.
❖IDE (Integrated Drive Electronics) 100 - 133 MB/s ❖Before you can store files in a partition,
you must format it with a filesystem.
❖Small Computer Systems Interface (SCSI) 320 MB/s

2: 300 MB/s ➢Filesystems provide a structure that specifies how data should reside
❖Serial Advanced Technology Attachment (SATA) 3: 600 MB/s on the storage device itself.
➢Filesystems also define storage-related features that operating
2: 6 GB/s
5: 45 GB/s (being systems can use, including permissions, compression, encryption,
❖Serial Attached SCSI (SAS) 3: 12 GB/s
developed) and journaling.
4: 22.5 GB/s (2017)
o A journaling filesystem uses a journal to keep track of the
❖Non-Volatile Memory Express (NVMe) 3.5 GB/s information written to the storage device.

5/63
/50 6/63
/50

Partitions, Filesystems, and Volumes Partitions, Filesystems, and Volumes

❖If you copy a file on the filesystem from one folder to another, that file ❖If the power is turned off during this process, information might not be
must pass into memory and then be written to the new location. transmitted as expected and data might be lost or corrupted.

Copy

RAM

7/63
/50 8/63
/50
 Partitions, Filesystems, and Volumes Partitions, Filesystems, and Volumes
❖With a journaling filesystem, each step required to copy the file to the ❖A filesystem may reside on one partition on your storage device or span
new location is first written to a journal. multiple partitions on different storage devices using RAID.
➢This allows the system to automatically retrace the steps taken prior
to a power outage and complete the file copy, after the power is
restored.

+ +

❖NTFS, ReFS are journaling filesystems, while FAT32 and exFAT are not. o X: is called a simple volume. C: is called a RAID volume.
9/63
/50 10/63
/50

Partitions, Filesystems, and Volumes Partitions, Filesystems, and Volumes

❖Since the introduction of MS-DOS on the IBM PC, the A: and B: drive ❖Rather than using a drive letter to represent a filesystem, volumes can
letters have been reserved for representing filesystems on floppy disks. instead use an existing folder to represent a filesystem.
➢No longer used on modern systems, their drive letters still remain ➢Use the C:\Data folder to represent a filesystem on a partition
reserved. o Users can access the folders and files on the filesystem by
navigating to the C:\Data folder.

➢UNIX and Linux systems use

this structure.

❖The term mounting refers to the process of associating a drive letter

or folder to a filesystem.
Write speed: 250 kbps Capacity: 800 KB to 2.8 MB
11/63
/50 12/63
/50
 Partition Types and Strategies Partition Types and Strategies
❖Besides partition mounted to C:, the Windows Server installation ❖The recovery and UEFI system partitions are not made accessible to the
program creates a recovery partition formatted with NTFS. operating system via a drive letter or folder, they are not considered
➢Contains tools you can use to repair a Windows Server installation volumes.
that is unable to boot.

❖On systems that use a UEFI, the installation program also creates a
UEFI system partition (also called an EFI system partition) formatted
with FAT32.
➢Store the boot loader program used to load the operating system
after the system is powered on.
❖On Microsoft systems
❑ ➢Partition that contains the boot loader is called system partition.
✓
➢Partition that contains operating system is called boot partition.
13 14/63
/50

Partition Types and Strategies Partition Types and Strategies

❖The table of all partition information for a storage device is located at ❖On systems that have a standard (non-UEFI) BIOS, the boot loader is
the beginning of the storage device itself. stored on the recovery partition, and the recovery partition is assigned
an active label.
➢The traditional partition table is called
the Master Boot Record (MBR),
➢The standard BIOS will search this active partition for an operating
and is limited to storage devices that
system boot loader after the system is powered on.
are equal to or less than 2 TB in size.

➢Newer storage devices and storage

devices larger than 2 TB use a GUID
Partition Table (GPT).

➢The MBR and GPT are functionally equivalent

❑
15/63
/50 16/63
/50
 Partition Types and Strategies Partition Types and Strategies
❖Example MBR partition layout for the first storage device on a system ❖Example GPT partition layout for the first storage device on a system
with a standard BIOS with a UEFI BIOS

17/63
/50 18/63
/50

Partition Types and Strategies Partition Types and Strategies

❖Many server administrators will create additional volumes on storage Advantage of multiple volume:
devices. ❖Segregate different types of data will make locating the log files easier.
❖Ensure that they cannot consume all available space on the C: volume,
which slow or halt the performance of the operating system.
❖Allow for the use of more than one type of filesystem on a single
storage device (NTFS, ReFS, etc.)
❖Reduce the chance that filesystem corruption will render all data on the
storage device unusable.
❖Speed up access to stored data by keeping filesystems as small as
❖Many server administrators prefer to configure a single volume on each possible.
storage device that uses the maximum available capacity.
➢Searching and maneuvering is much quicker and easier in a smaller,
defined space than in a larger one.

19/63
/50 20/63
/50
 RAID Types and Strategies RAID Types and Strategies
❖To protect data against storage device failure, you can spread ❖RAID level 0 (striping) divides an individual file
filesystems across several storage devices using a fault tolerant RAID into sections and saves it concurrently on two or
volume. more storage devices, one section per storage
❖In addition, to speed up access to storage devices and to combine device.
multiple storage devices into a single volume.
❖Striping is useful when you need to speed up
storage device access (e.g., with hard disks), but
it is not fault tolerant.
❖Fault-tolerant RAID volumes are not a replacement for performing
➢If one storage device fails in a striping
regular data backups.
configuration, all data is lost.
➢If files become corrupted, or are inadvertently deleted from a fault-
tolerant RAID volume, you will need to restore those files from a
backup.

21/63
/50 22/63
/50

RAID Types and Strategies RAID Types and Strategies

❖RAID level 1 (mirroring) provides fault tolerance ❖RAID level 3 is striping with a parity
in the event of a storage device failure. bit, which indicates what data is where.

❖The same data is written to two separate storage ❖Requires a minimum of three storage
devices at the same time. If one fails, the copy devices.
automatically replaces the failed storage device.
❖One of the storage devices used to store
❖Drawback to RAID level 1 is the cost. the parity information.

❖Can replace the storage device and regenerate the data using the
❖RAID level 2 is no longer used and was a variant of RAID 0 that parity information.
allowed for error and integrity checking on storage devices. Modern ➢If the parity storage device fails, the system must be restored from
storage devices do this intrinsically. a backup device.
23/63
/50 24/63
/50
 RAID Types and Strategies RAID Types and Strategies
❖RAID level 4 is only a slight variant on RAID level 3. ❖RAID level 5 replaces RAID levels 3 and 4.
❖Requires a minimum of three storage devices.
➢Offers greater access speed than RAID level 3 because it can store ❖The parity information is intermixed with data on the storage devices
data in blocks and, thus, does not need to access all storage devices that make up the set.
at once to read data. ➢This offers better performance and fault tolerance;
❖If two storage devices fail at the same time, the data must be restored
from a backup copy.

25/63
/50 26/63
/50

RAID Types and Strategies RAID Types and Strategies

❖RAID level 6 is fundamentally the same as RAID level 5, but it adds a
second set of parity bits for added fault tolerance.
❖Allows up to two simultaneous storage device failures while remaining
fault tolerant.
❖Requires a minimum of four storage devices,
➢But at least seven storage devices are required for acceptable
performance.

❖22 + 12 = 34
❖68 – 65 = 3
❖13 -9 = 4

27/63
/50 28/63
/50
 RAID Types and Strategies RAID Types and Strategies
❖RAID levels are often combined ❖RAID configurations can be handled by:
➢Software running on an operating system (called software RAID)
➢RAID level 10 refers to a striping ➢Hardware contained on a storage controller (called hardware
configuration (RAID level 0) that is mirrored RAID)
(RAID level 1) to another striping configuration. ➢System BIOS (called firmware RAID).

❖To ensure that the processor performance of the system is not impacted
➢RAID level 15 refers to a striping with following a RAID level 5 storage device failure, storage controllers that
parity configuration (RAID level 5) that provide hardware RAID level 5 contain a small processor that is used to
is mirrored (RAID level 1) to another perform the calculations needed to provide data to users.
striping with parity configuration.
❖Most firmware RAID devices only support RAID levels 0 and 1.

29/63
/50 30/63
/50

SAN Storage
❖Some modern rackmount servers contain only enough storage to host
the operating system and associated server programs.

❖All other data files,

databases, Web content,
virtual hard disk files, and
2 SAN Storage so on are stored on a high-
capacity external storage
area network (SAN)
device via a SAN protocol.

31 32/63
/50
 SAN Storage SAN Storage
❖SAN devices use hardware RAID internally to provide fault tolerance. ❖SAN devices are not the same as Network-Attached Storage (NAS)
devices.
❖SAN devices only provide storage to other systems via SAN protocol.
❖NAS devices often contain several storage devices configured using
hardware RAID.
❖Do not format or manage filesystems on this storage.
➢Instead, the operating system that connects to the SAN device
provides this functionality. ❖NAS devices run an operating
➢As a result, it is functionally similar to an external USB drive system and function as a file
server on the network that
clients connect to using SMB,
NFS, or File Transfer Protocol
(FTP).

33/63
/50 34/63
/50

iSCSI SAN Device iSCSI SAN Device

❖Internet Small Computer Systems Interface (iSCSI) is a transfer ❖A single iSCSI target is normally composed of multiple storage devices
protocol that uses Ethernet network cables to transfer data to and from combined together using fault tolerant hardware RAID on the SAN
a SAN device, either on the local network or across the Internet, using device itself.
the SCSI protocol at speeds of up to 40 Gb/s.
❖As a result, the iSCSI initiator and operating system will see the iSCSI
❖The software component in target as a single storage device.
the operating system that
connects to the SAN device
via iSCSI is referred to as an ❖You can then create volumes on this storage device as you would on
iSCSI initiator any other storage device.
❖The storage on the SAN
device that is made available ❖A single iSCSI SAN device can be accessed by multiple servers.
to the iSCSI initiator is called
the iSCSI target.
35/63
/50 36/63
/50
 iSCSI SAN Device iSCSI SAN Device
❖After you have connected your system to the iSCSI target, the iSCSI ❖You can configure the iSCSI
target should be identified by the Windows Server system as a new initiator on Windows Server
storage device that you can create volumes on in Server Manager or the running in a Hyper-V virtual
Disk Management tool. machine (VM) to connect to an
iSCSI target.
❖Most server administrators create simple volumes on storage devices
that are provided by iSCSI targets, because the iSCSI target uses fault ❖Must ensure that the VM is
tolerant hardware RAID on the underlying storage devices. connected to an external
➢However, if your system connects to multiple iSCSI targets, you can virtual switch associated with
create software RAID volumes from the storage provided by the the iSCSI-compliant network
iSCSI targets for added fault tolerance. interface on your rackmount
server that has a physical
connection to the iSCSI SAN
device.
37/63
/50 38/63
/50

iSCSI SAN Device Fibre Channel SAN Device

❖Some rackmount servers include an iSCSI initiator in the system BIOS. ❖Fibre Channel (FC) is a technology that can be used to transport
SCSI data to a remote FC SAN device, across an Ethernet or fiber optic
cable at speeds of up to 128 Gb/s.
❖In this case, you can configure the iSCSI initiator in the system BIOS to
connect to the iSCSI target before installing Windows Server. ❖A rackmount server uses a hardware-based FC controller called an FC
Host Bus Adapter (HBA) that connects to an FC SAN device.
➢This allows you to create the
system and boot partitions on
the iSCSI target during the
Window Server installation,
eliminating the need for local
storage devices in the
rackmount server.

39/63
/50 40/63
/50
 Fibre Channel SAN Device Connecting Windows Server to a Fibre Channel SAN Device

❖Each FC SAN device contains FC-capable storage devices connected via ❖An FC HBA can also be an
an FC switch. FC-capable 10 Gb Ethernet
network interface if the FC
SAN device uses FC over
Ethernet (FCoE).

❖FC-capable storage devices are normally SAS hard disks or SSDs that
contain additional FC circuitry, or NVMe SSDs that have an FC capable
memory controller.

41/63
/50 42

Using MPIO to Connect to Multiple SAN Devices Using MPIO to Connect to Multiple SAN Devices
❖Data center server environments ❖Larger data centers often have multiple
often have several iSCSI or FC SAN SANs that host the same information.
devices.
❖In this case, servers can have multiple
❖Each server in a data center can connections to different SANs to provide
have multiple connections to a fault tolerance in case a single SAN
single SAN for fault tolerance in becomes unavailable, or to load balance
case a single connection becomes requests across multiple SANs for greater
unavailable, or to load balance speed.
requests across multiple
connections for greater speed.
❖These configurations are called Multipath
Input output (MPIO).

43/63
/50 44/63
/50
 Enabling Data Deduplication
❖Data Deduplication
➢Users may store multiple copies of the same document in different
folders on the same volume. → Wasting unnecessary space.

3 Managing Volume Data

45 46/63
/50

Enabling Data Deduplication Enabling Data Deduplication

❖Data Deduplication ❖Data Deduplication
➢Data Deduplication Service on Windows Server will periodically
scan for duplicate file contents that have not been modified for a ➢Data deduplication ignores file names when scanning for files to
certain number of days. deduplicate.
o This allows files with the same contents to be deduplicated, even
o If duplicate file contents are found, the underlying filesystem will if they have different file names.
store a single copy of those contents and ensure that each file
name refers to that single copy.
➢Windows Server allows you to enable data deduplication for NTFS
o If a user modifies the contents of a file that has been and ReFS volumes that are not the boot or system volume.
deduplicated, a new copy of the modified file is created and
associated with the particular file name.

47/63
/50 48/63
/50
 Optimizing Volumes Optimizing Volumes
❖When saving a file to a filesystem on a hard disk, ❖When accessing a fragmented file
➢OS saves the file to the first area of available space without caring ➢Slowing access time and creating additional hard disk wear.
the size of that area.
o Might be saved to several different areas of the hard disk.

❖Defragmentation: defragmenting
locates fragmented files and moves
them to a location on the hard disk
where they can be stored and read
(Fragmented File) as a single unit.
49/63
/50 50/63
/50

Optimizing Volumes Optimizing Volumes

❖Defragmenting an SSD is called trimming
❖Because SSDs use non-volatile
➢Erases sections of the non-volatile memory chips on the SSD that
memory chips for storage, they do not
no longer contain data, consolidating the existing data into fewer
experience the same type of
memory locations.
fragmentation that hard disks do.

❖When saving a file, it is saved to a new memory location, even if a

previous memory location is no longer used (previous file was deleted).
➢This is because SSD storage controllers do not automatically remove
data from unused memory locations in order to maintain fast
performance.
o SSD keeps track of more memory addresses than it stores data
for → slowing access time.
51/63
/50 52/63
/50
 Repairing Volumes Repairing Volumes
❖As the number of applications and users that access and ❖If users report errors, you should check for, and repair, errors on the
modify data on a volume increases, so does the volume, such as corrupted data or bad sectors.
possibility of storage subsystem errors that result in
corrupted data or invalid filesystem information.

❖An area of the hard disk can no longer

hold a magnetic charge (called a bad
sector); thus, cannot be written to by
the storage subsystem.

❖When bad sectors are identified by the storage subsystem, their

locations are recorded in a table of bad sectors on the filesystem to
ensure that they are no longer used to store data.
53/63
/50 54/63
/50

Repairing Volumes Backing Up and Restoring Data

❖The command provides the most functionality and can be run ❖While fault tolerant RAID volumes protect against storage device
at a Command Prompt or Windows PowerShell window. failure, they do not protect data when:
➢ . ➢One or more files are deleted or become corrupted
➢A filesystem becomes corrupted
➢A server fails, and the volumes hosted by the server become
unavailable

❖The most reliable way to protect data is

to back the data up on a regular basis
(back up).
❖ The ReFS filesystem prevents corrupted data from being stored. As a
result, you cannot check a volume formatted with ReFS for errors
55/63
/50 56/63
/50
 Backing Up and Restoring Data Backing Up and Restoring Data
❖When you back up data, a copy is made to ❖Full backup: back up all of the data that you specify, regardless of
another folder on the system or on another system whether the files have been modified since the last backup.
across the network. ➢May take a long time to complete, depending on the amount of data
➢If the original data becomes unavailable, it can that is backed up.
be restored from the backup copy.

❖Incremental backup:
➢At first time, full backup.
❖Windows Server Backup can use to back up files on NTFS and ReFS
➢From the next time, only back up data
volumes to a dedicated storage device, local volume, or shared folder on
that has been modified since the previous backup.
the network.
o Less time to complete.
➢To restore data, you must restore the first full backup, followed by
❖Many organizations purchase third-party backup software. all subsequent incremental backups, in order.
57/63
/50 58/63
/50

Backing Up and Restoring Data Backing Up and Restoring Data

❖Differential backup also only saves the added and changed files in ❖Notes:
the backup, but it is based on the latest full backup. ➢Each full backup performed by the Windows Server Backup tool
overwrites the previous full backup.

➢Windows Server Backup will perform a maximum of 14 incremental

backups after a full backup.
o After 14 incremental backups, another full backup is performed,
followed by another 14 incremental backups, and so on.

❖Can choose to back up entire volumes, certain folders or files on a

volume, or the contents of the recovery or EFI system partitions.
59/63
/50 60/63
/50
 Backing Up and Restoring Data Backing Up and Restoring Data
❖Additionally, you can choose: ❖By default, the Volume Shadow Copy Service (VSS) is used to back up
➢Bare metal recovery—This creates a full backup of the entire files that users currently have open by taking a copy of those files that
system. are included in the backup.
o To restore this backup, you must start a Windows Server
installation, click Repair your computer, and select the Restore
your computer using a system image that you created earlier
option.
➢System state—This creates a backup of all operating system
settings, including the Windows Registry, Active Directory database
(for domain controllers), and boot loader files.
➢Hyper-V—This backs up the configuration and virtual hard disk files
for the virtual machines you select and is only available if the Hyper-
V role is installed.

61/63
/50 62/63
/50

THANK YOU FOR YOUR ATTENTION

Nguyen Viet Ha, Ph.D.

Department of Telecommunications and Networks
Faculty of Electronics and Communications
The University of Science, Vietnam National University, Ho Chi Minh City
Email: nvha@hcmus.edu.vn