TI2122 - Komputasi Awan 09
TI2122 - Komputasi Awan 09
TI2122 - Komputasi Awan 09
Foreword
• Data has become an enterprise’s most important asset. How is data stored on the
cloud? How is this different from the way it is stored with traditional IT? This chapter
will answer these questions.
Course Objectives
• Upon completion of this course, you will:
• Understand Huawei’s FusionSphere virtualization solution.
• Understand the storage architecture used for virtualized environments.
• Understand different types of disks.
• Understand the differences between centralized and distributed storage.
• Understand the differences between virtualized and non-virtualized storage.
• Understand different types of VM disks.
• Understand the storage features of Huawei’s virtualization product.
Course Contents
• Storage Architecture for Virtualization
• Physical Disk Types and Related Techniques
• Centralized Storage vs. Distributed Storage
• Virtualized Storage vs. Non-Virtualized Storage
• Introduction to VM Disks
• Storage Features of Huawei’s Virtualization Product
Storage Virtualization
• Storage virtualization is a significant component of storage servers & facilitates the
management and monitoring of storage in a virtualized environment.
• Storage virtualization basically combines/pools the storage that is available in various
devices and keeps it as single storage. Identification of the available storage is done by
leveraging the software and aggregating them to use it in a virtual system/environment.
• This logical storage will look like physical storage to a server
or host, which means server won’t be able to differentiate
between physical storage and logical storage. In storage
virtualization, multiple physical disks are combined into a group
and from that group of physical disks, virtual storage or logical
storage blocks are assigned to a server for use.
Benefit of Storage Virtualization
• It is highly scalable
• It allows easy addition and deletion of storage without affecting any application
• Easy data migration
• Easy storage management
Storage Virtualization
• https://www.youtube.com/watch?v=O0D8Ftc44ls
• https://www.youtube.com/watch?v=igHywP9VoNk
Importance of Storage Virtualization
• Storage Tiering: Using the storage technique as a bridge or as a stepping stone, this
technique analyzes and selects the most commonly used data & places it in its
highest-performing storage pool and the least-used data in the weakest-performance
storage pool.
• WAN Environment: Instead of sending multiple copies of the same data over WAN,
the WAN accelerator is used to locally cache the data and present it at a LAN speed,
and not impact the WAN performance.
• SAN Storage: SAN technology presents the storage as block-level storage & the
storage is presented over the Ethernet network of OS.
• File Server: OS writes the data to a remote server location to keep it separate and
secure from local users.
Type of Storage Virtualization
Type of Storage Virtualization
• 1. Block Virtualization
• In block virtualization, we basically separate our logical storage from that of the
physical so that the user/administrator can access without having to access the
physical storage, basically doing this way helps the administrator in giving a lot of
flexibility in managing different storage.
• 2. File Virtualization
• In File virtualization, it basically removes the dependencies caused in accessing the
data at file level to that of the location where they are actually present. This basically
helps in overcoming the challenges faced with network-attached storage and they
also help in optimizing the storage usage and also help us to do some file migrations
in a non-disruptive way.
https://www.youtube.com/watch?v=5EqAXnNm0FE
Configuration of Storage Virtualizations
Configuration of Storage Virtualizations
• 1. In-Band Approach (Symmetric)
• In this method, we store the virtual environment configuration in the data path itself
as in the data as well as the control flow. This kind of solution is considered
easy/simple to implement as we do not use any kind of software. We do different
levels of abstraction inside the data path. These kinds of solutions help us to improve
our device’s performance majorly and also prolong the useful life of the devices. One
of the examples of an in-band based solution is that IBM’s total storage area network
volume controller.
• 2. Out-Band Approach (Asymmetric)
• In this approach, the implementation of the virtual environment is done outside of the
data path as in the data flow and the control flow are separated which can be
achieved by separating our Metadata from data and putting them in different places.
This kind of virtualization involves in transferring all the tables to a Metadata
controller which has all the Metadata files. By separating both the flows we achieve
the usage of complete bandwidth that is offered by the storage area network.
Implementation Storage Virtualization
• https://www.youtube.com/watch?v=QZntIjLz-kA
Where Do the Storage Resources Used by
VMs Come From?
How do VMs access
the underlying
storage resources?
?
Storage Architecture for Virtualization
Non-
Virtualized
virtualized
storage
storage
Shared Catalog
File System
Logical
Volumes
Distributed
Storage
Virtualized Storage Conversion Path in Cloud
Computing
Shared Catalog
Virtual file
system
Logical
volume
Attach to compute cluster
and format
Physical
NFS file
volume Logical division
Format system
RAID or Replication
Non-Virtualized Storage Conversion Path in
Cloud Computing
Logical
volume Attach to compute cluster
Physical
volume Logical division
RAID or Replication
Next Contents
• Storage Architecture for Virtualization
• Physical Disk Types and Related Techniques
• Centralized Storage vs. Distributed Storage
• Virtualized Storage vs. Non-Virtualized Storage
• Introduction to VM Disks
• Storage Features of Huawei’s Virtualization Product
Physical Disk Type - SATA
• SATA (Serial Advanced Technology
Attachment), also called serial ATA, is a
computer bus interface that connects host
bus adapters to mass storage devices.
Using serial communication, the serial
ATA bus uses embedded clock signals to
check transmission commands (not only
data) and automatically corrects detected
errors, ensuring reliable data
transmission for SATA disks. The SATA
interface is simple and the disks are hot-
swappable
Physical Disk Type - SAS
• Serial Attached SCSI (SAS) is a new-
generation SCSI technology, which, like
the popular Serial ATA (SATA), also uses
serial communication to achieve a higher
transmission speed. Compared with the
older parallel SCSI, SAS improves the
efficiency, availability, and scalability of
storage systems and offers optional
compatibility with SATA
SAS Interface
Physical Disk Type - NL-SAS
• Nearline SAS (NL-SAS) disks are a
combination of a SAS interface with SATA
disks. The rotational speed of NL-SAS
disks is only 7200 rpm, so performance is
lower than that of SAS disks. However,
with the SAS interface, addressing and
speed are improved.
NL-SAS
Physical Disk Type - SSD
• A solid-state drive (SSD) is a solid-state
storage device that uses integrated circuit
assemblies as non-volatile memory. An
SSD consists of a control unit and a
storage unit (flash or DRAM chip). An SSD
is the same as a common hard disk drive
(HDD) in terms of interface specifications
and definition, functionality, usage, and
product shape and size. Although SSDs
deliver fast read/write performance, light
weight, low power consumption, and
compact sizes which traditional HDDs
cannot rival, SSDs have a shorter lifespan
and higher price.
Internal Structure of SSD
Advantages and Disadvantages
• High read speed • High cost
• Superior shock resistance • Limited services life
• No noise
• Small size and lightweight
• Wider operating temperature range
Comparing
Performance
Price
Comparison
ITEM SSD SAS NL-SAS HDD
Performance Very High High Relatively High Relatively High
Reliability Minor High Relatively High Minor
Price Very High High Relatively High Low
Power Minor High Relatively High Relatively Low
Consumption
Recommend Suitable for Suitable for Suitable for Suitable for
Scenario users with very high- and mid- users with users with
large data end users with large data large data
access discrete data blocks and low blocks and low
service I/O service
pressure pressure
Question & Answers