Introduction to storage virtualization
This chapter defines the concept of storage virtualization and provides an overview about how to apply virtualization to address the challenges of storage requirements. It includes the following topics:
1.1 Storage virtualization terminology
Storage virtualization is a term that is used extensively throughout the storage industry. It can be applied to various technologies and underlying capabilities. In reality, most storage devices technically can claim to be virtualized in one form or another. Therefore, this chapter starts by defining the concept of storage virtualization as it is used in this book.
IBM describes storage virtualization in the following way:
•Storage virtualization is a technology that makes one set of resources resemble another set of resources, preferably with more desirable characteristics.
•It is a logical representation of resources that is not constrained by physical limitations and hides part of the complexity. It also adds or integrates new function with existing services and can be nested or applied to multiple layers of a system.
When the term storage virtualization is mentioned, it is important to understand that virtualization can be implemented at various layers within the I/O stack. There must be a clear distinction between virtualization at the disk layer (block-based) and virtualization at the file system layer (file-based).
The focus of this publication is virtualization at the disk layer, which is referred to as block-level virtualization or the block aggregation layer. A description of file system virtualization is beyond the intended scope of this book.
The Storage Networking Industry Association’s (SNIA) block aggregation model provides a useful overview of the storage domain and the layers, as shown in Figure 1-1. It illustrates several layers of a storage domain:
•File
•Block aggregation
•Block subsystem layers
Figure 1-1 SNIA block aggregation model1
The model splits the block aggregation layer into three sublayers. Block aggregation can be realized within hosts (servers), in the storage network (storage routers and storage controllers), or in storage devices (intelligent disk arrays).
The IBM implementation of a block aggregation solution is IBM Spectrum Virtualize™ software, running on IBM SAN Volume Controller and the IBM Storwize family.
The IBM SAN Volume Controller is implemented as a clustered appliance in the storage network layer. The IBM Storwize family is deployed as modular storage that provides capabilities to virtualize its own internal storage and external storage.
The key concept of virtualization is to decouple the storage from the storage functions that are required in the storage area network (SAN) environment.
Decoupling means abstracting the physical location of data from the logical representation of the data. The virtualization engine presents logical entities to the user and internally manages the process of mapping these entities to the actual location of the physical storage.
The actual mapping that is performed depends on the specific implementation, such as the granularity of the mapping, which can range from a small fraction of a physical disk up to the full capacity of a physical disk. A single block of information in this environment is identified by its logical unit number (LUN), which is the physical disk, and an offset within that LUN, which is known as a logical block address (LBA).
The term physical disk is used in this context to describe a piece of storage that might be carved out of a Redundant Array of Independent Disks (RAID) array in the underlying disk subsystem. Specific to the IBM Spectrum Virtualize implementation, the address space that is mapped between the logical entity is referred to as a volume. The array of physical disks is referred to as managed disks (MDisks).
Figure 1-2 shows an overview of block-level virtualization.
Figure 1-2 Block-level virtualization overview
The server and application are aware of the logical entities only, They access these entities by using a consistent interface that is provided by the virtualization layer.
The functionality of a volume that is presented to a server, such as expanding or reducing the size of a volume, mirroring a volume, creating an IBM FlashCopy®, and thin provisioning, is implemented in the virtualization layer. It does not rely in any way on the functionality that is provided by the underlying disk subsystem. Data that is stored in a virtualized environment is stored in a location-independent way, which enables a user to move or migrate data between physical locations, which are referred to as storage pools.
IBM Spectrum Virtualize provides the following benefits:
•Online volume migration while applications are running, which is possibly the greatest single benefit for storage virtualization. This capability enables data to be migrated on and between the underlying storage subsystems without any effect on the servers and applications. In fact, this migration is performed without the knowledge of the servers and applications that it even occurred.
•Simplified storage management by providing a single image for multiple controllers, and a consistent user interface for provisioning heterogeneous storage.
•Enterprise-level Copy Services functions. Performing Copy Services functions within the IBM SAN Volume Controller, removes dependencies on the storage subsystems. Therefore, it enables the source and target copies to be on other storage subsystem types.
•Storage usage can be increased by pooling storage across the SAN.
•System performance is often improved with IBM Spectrum Virtualize and IBM SAN Volume Controller as a result of volume striping across multiple arrays or controllers and the other cache that it provides.
•Software-based encryption capabilities to provide improved data security among storage virtualization solutions.
•Data replication to cloud storage using advanced copy services for data migration and backup solutions.
•Data reduction techniques for space efficiency such as Thin Provisioning and compression.
The IBM SAN Volume Controller delivers all these functions in a homogeneous way on a scalable and high availability software platform over any attached storage and to any attached server.
1.2 Requirements driving storage virtualization
Today, many organizations are searching for affordable and efficient ways to store, use, protect, and manage the data. An emphasis is on the IBM Cognitive era of clients’ businesses and their dynamic infrastructure. Therefore, a storage environment requires an easy to manage interface and flexibility to support many applications, servers, and mobility requirements.
Business demands change quickly.
The following key client concerns drive storage virtualization:
•Growth in data center costs
•Inability of IT organizations to respond quickly to business demands
•Poor asset usage
•Poor availability or service levels
•Lack of skilled staff for storage administration
You can see the importance of addressing the complexity of managing storage networks by applying the total cost of ownership (TCO) metric to storage networks. Industry analyses show that storage acquisition costs are only about 20% of the TCO. Most of the remaining costs relate to managing the storage system.
But how much of the management of multiple systems, with separate interfaces, can be handled as a single entity? In a non-virtualized storage environment, every system is an “island” that must be managed separately.
1.2.1 Benefits of using IBM Spectrum Virtualize
IBM SAN Volume Controller running IBM Spectrum Virtualize software reduces the number of separate environments that must be managed down to a single environment. It also provides a single interface for storage management and various functions. After the initial configuration of the storage subsystems, all of the day-to-day storage management operations are performed by using the graphical user interface of the IBM Spectrum Virtualize.
Because IBM Spectrum Virtualize provides many functions, such as mirroring and IBM FlashCopy, there is no need to acquire additional subsets of applications for each attached disk subsystem that is virtualized by IBM Spectrum Virtualize.
Today, it is typical that open systems run at less than 50% of the usable capacity that is provided by the RAID disk subsystems. A block-level virtualization solution, such as IBM Spectrum Virtualize, can allow significant savings, increase effective capacity of storage systems up to five times, and decreases your need for floor space, power, and cooling.
1.3 Latest changes and enhancements
IBM Spectrum Virtualize V8.1 and its related hardware upgrade represents an important milestone in the product line development, with important enhancements. Because of the recent changes, the internal architecture of the code IBM Spectrum Virtualize is significantly rebuilt, enabling the system to break the previous limitations in terms of scalability, flexibility, and functionality.
The intent of this book is to cover the major software features and provide a brief summary of supported hardware.
1.3.1 IBM SAN Volume Controller Engine 2145-SV1
The IBM SAN Volume Controller node model 2145-SV1 is also based on IBM System x server technology. Each of these models is delivered in a 2U 19-inch, rack-mount enclosure.
The node model 2145-SV1 features two Intel Xeon E5 v4 eight-core processors and 64 gigabytes (GB) of memory with options to increase the total amount of memory up to 256 GB.
The IBM SAN Volume Controller model 2145-SV1 features three 10-gigabit (Gb) Ethernet ports that can be configured as iSCSI connectivity and system management. The node model 2145-SV1 supports up to four I/O ports adapters that provide up to sixteen 16 Gb Fibre Channel ports or up to four 10 Gb Ethernet ports to be configured as iSCSI or Fibre Channel over Ethernet (FCoE).
In a clustered solution, the IBM SAN Volume Controller node model 2145-SV1 can support up to 20 expansion enclosures.
Figure 1-3 Front view of 2145-SV1
The 2145-SV1 model is complemented with the following expansion enclosures (attachable also to the 2145-DH8):
•2145-12F holds up to 12 LFF 3.5” SAS-attached disk drives in 2U enclosures
•2145-24F accommodates up to 24 SFF 2.5” SAS-attached disks in 2U enclosures
•2145-92F adds up to 92 internal 3.5” SAS flash drives in 5U enclosures
|
Note: Model 2147 (including expansion enclosures) has identical hardware to Model 2145, but delivered with the enterprise-level remote IBM support that offers these features:
•Technical Advisors to proactively improve problem determination and communication
•On-site and remote software installation and updates
•Configuration support
•Enhanced response times for high severity problems
|
1.3.2 IBM Spectrum Virtualize Software changes and enhancements
At the V7.3 announcement, IBM included significant upgrades of the 2145-DH8 hardware. The 2145-DH8 introduced a 2U server based on the IBM x3650 M4 series and integrates the following features:
•Minimum eight-core processors with 32 GB memory for the IBM SAN Volume Controller. Optional secondary processor with extra 32 GB memory when a third I/O card is needed. Secondary processor is compulsory when IBM Real-time Compression™ is enabled.
•Integrated dual-battery pack as an uninterruptible power supply (UPS) in a power outage. An external UPS device is no longer needed, avoiding miscabling issues.
•Dual, redundant power supplies, so no external power switch is required.
•Removed front panel. Most of its actions were moved to the functionality of the rear Technician Port (Ethernet port) with enabled Dynamic Host Configuration Protocol (DHCP) for instant access.
•Two boot drives with data mirrored across drives. The SVC node will still boot in a drive failure. Dump data is striped for performance reasons.
•Enhanced scalability with up to three Peripheral Component Interconnect Express (PCIe) slot capabilities, which allow users to install up to three 4-port 8 Gbps FC host bus adapters (HBAs), for a total of 12 ports. It supports one four-port 10 gigabit Ethernet (GbE) card (iSCSI or FCoE) and one dual-port 12 Gbps serial-attached SCSI (SAS) card for flash drive expansion unit attachment (model 2145-24F).
•Improved Random Access Compression Engine (RACE) with the processing offloaded to the secondary dedicated processor and using 36 GB of dedicated memory cache. At a minimum, one Compression Accelerator card needs to be installed (up to 200 compressed volumes) or two Compression Accelerators allow up to 512 compressed volumes.
•Optional 2U expansion enclosure 2145-24F with up to 24 flash drives (200, 400, 800, or 1600 GB).
•Extended functionality of IBM Easy Tier® by storage pool balancing mode within the same tier. It moves or exchanges extents between highly utilized and low-utilized MDisks within a storage pool, increasing the read and write performance of the volumes. This function is enabled automatically in IBM SAN Volume Controller, and does not need any licenses.
•The SVC cache rearchitecture splits the original single cache into upper and lower caches of different sizes. Upper cache uses up to 256 megabytes (MB), and lower cache uses up to 64 GB of installed memory allocated to both processors (if installed). Also, 36 GB of memory is always allocated for Real-time Compression if enabled.
•Near-instant prepare for FlashCopy due to the presence of the lower cache. Multiple snapshots of the golden image now share cache data (rather than several N copies).
For V7.4, IBM announced the following changes:
•Hardware change:
The SVC introduces a 16 Gbps FC adapter based on the Emulex Lancer multiprotocol chip, which offers either FC connectivity or FCoE to the SVC. The adapter can be configured as a two-port 16 Gbps FC, four-port 8 Gbps FC, or four-port 10 GbE profile.
•Software changes:
– The most noticeable change in V7.4 after the first login is the modified graphical user interface (GUI) with the new layout of the system window, including enhanced functions that are available directly from the welcome window.
– The concept of the GUI design conforms to the well-known approach from IBM System Storage XIV® Gen3 and IBM FlashSystem® 840. It provides common, unified procedures to manage all these systems in a similar way, enabling administrators to simplify their operational procedures across all systems.
– Child pools are new objects, which are created from the physical storage pool and provide most of the functions of managed disk groups (MDiskgrps), for example, volume creation. However, the user can specify the capacity of the child pool at creation.
– In previous SVC systems, the disk space of a storage pool was from MDisks, so the capacity of a storage pool depended on the capacity of the MDisks. The user could not freely create a storage pool with a particular capacity that they wanted. The maximum number of storage pools remains at 128, and each storage pool can have up to 127 child pools. Child pools can only be created in the command-line interface (CLI), but they are shown as child pools with all their differences to parent pools in the GUI.
– A new level of volume protection prevents users from removing mappings of volumes that are considered active. Active means that the system has detected recent I/O activity to the volume from any host within a protection period that is defined by the user. This behavior is enabled by system-wide policy settings. The detailed volume view contains the new field that indicates when the volume was last accessed.
– A user can replace a failed flash drive by removing it from the 2145-24F expansion unit and installing a new replacement drive, without requiring a Directed Maintenance Procedure (DMP) to supervise the action.
– The user determines that the fault light-emitting diode (LED) is illuminated for a drive, so they can expect to be able to reseat or replace the drive in that slot. The system automatically performs the drive hardware validation tests and promotes the unit into the configuration if these checks pass.
– The 2145-DH8 with IBM SAN Volume Controller software version 7.4 and later supports the T10 Data Integrity Field between the internal RAID layer and the drives that are attached to supported enclosures.
– The SVC supports 4096-bytes native drive block size without requiring clients to change their block size. The SVC supports an intermix of 512 and 4096 drive native block sizes within an array. The GUI recognizes drives with different block sizes and represents them with different classes.
– The IBM SAN Volume Controller 2145-DH8 improves the performance of Real-time Compression and provides up to double I/O operations per second (IOPS) on the model DH8 when it is equipped with both Compression Accelerator cards. It introduces two separate software compression engines (RACE), taking advantage of multi-core controller architecture. Hardware resources are shared between both RACE engines.
– Adds virtual local area network (VLAN) support for iSCSI and Internet Protocol (IP) replication. When VLAN is enabled and its ID is configured for the IP addresses that are used for either iSCSI host attach or IP replication on the SVC, appropriate VLAN settings on the Ethernet network and servers must also be correctly configured to avoid connectivity issues. After the VLANs are configured, changes to their settings disrupt the iSCSI or IP replication traffic to and from the SVC.
– New informational fields are added to the CLI output of the lsportip, lsportfc, and lsportsas commands, indicating the physical port locations of each logical port in the system.
At the V7.5 announcement, IBM included the following software changes:
•Direct host attachment through 16 Gbps FC adapters to all operating systems except IBM AIX.
•Support for Microsoft Offloaded Data Transfer (ODX).
•Introduction of IBM HyperSwap® topology, which enables each volume to be presented by two I/O groups by two pairs of SVC nodes. The configuration tolerates combinations of node and site failures by using a flexible choice of host multipathing driver interoperability.
•Support of VMware vSphere V6.0 virtual volumes (VVols). Each virtual machine (VM) keeps different types of data each in a VVol, each of which is presented as a volume (logical unit is SCSI) by the IBM Spectrum Virtualize system. Therefore, each VM owns a small number of volumes.
At the V7.6 announcement, IBM included the following changes:
•Hardware changes:
– The IBM SAN Volume Controller model 2145-DH8 supports up to four quad-port 16 Gbps native FC adapters, either for SAN access or direct host attachment.
– Removed support of IBM SAN Volume Controller modes 2145-8G4 and 8A4.
•Software changes:
– Visual and functional enhancements in the GUI, with changed menu layout and an integrated performance meter on main page.
– Implementation of Distributed RAID, which differs from traditional RAID arrays by eliminating dedicated spare drives. Spare capacity is spread across disks, making the reconstruction of failed disk faster.
– Introduced software encryption enabled by IBM Spectrum Virtualize and using AES256-XTS algorithm. Encryption is enabled on the storage pool level. All newly created volumes in such pool are automatically encrypted. An encryption license with Universal Serial Bus (USB) flash drives is required.
– Developed the Comprestimator tool, which is included in IBM Spectrum Virtualize software. It provides statistics to estimate potential storage savings. Available from the CLI, it does not need compression licenses and does not trigger any compression process. It uses the same estimation algorithm as an external host-based application, so results are similar.
– Enhanced GUI wizard for initial configuration of HyperSwap topology. IBM Spectrum Virtualize now allows IP-attached quorum disks in HyperSwap system configuration.
– Increased the maximum number of iSCSI hosts attached to the system to 2048 (512 host iSCSI qualified names (IQNs) per I/O group) with a maximum of four iSCSI sessions per SVC node (8 per I/O group).
– Improved and optimized read I/O performance in HyperSwap system configuration by parallel read from primary and secondary local volume copies. Both copies must be in a synchronized state.
– Extends the support of VVols. Using IBM Spectrum Virtualize, you can manage one-to-one partnership of VM drives to IBM SAN Volume Controller volumes. It eliminates single, shared volume (data store) I/O contention.
– Customizable login banner. Using CLI commands, you can now define and show a welcome message or important disclaimer on the login window to users. This banner is shown in GUI or CLI login window.
At V7.7, IBM announced the following changes:
•Software changes:
– New software capabilities to support encryption protection of data in DRAID volumes.
– Enhanced external virtualization flexibility through iSCSI protocol.
– IP Link compression algorithm to improve usage of IP networks. This feature reduces the volume of data that must be transmitted during remote copy operations.
– GUI support for IP Quorum to help administrators to configure IBM SAN Volume Controller as Stretched Cluster or HyperSwap solution using IP Quorum solution.
– GUI support for IBM Comprestimator Utility to assist administrators to use the compression functions to display thin provisioning and compression estimates analysis for single or multiple volumes.
– Support for up to 10,000 volumes per clustered system running code level 7.7.1.
•Hardware changes:
– Introduced new model type 2145-SV1 with improved system design. Includes model 2147 with enhanced enterprise-class IBM remote and onsite support.
– Host Port fabric virtualization (N_Port ID Virtualization (NPIV)) for Fibre Channel host attachment.
– Extended the cache for read operations to 64 GB.
– Up to two expansion enclosures are supported by SVC model DH8 and SV1 node pair or I/O group, delivering up to 24 hard disk drives as internal capacity.
– Support for large capacity Read Intensive SSDs.
At V7.8, IBM incorporated the following innovative changes:
•Software change:
– Support for Cloud Storage. IBM introduced software capabilities to IBM Spectrum Virtualize to interface with an external cloud storage service provider.
– Management GUI supports configuration and management of host clusters including management of mappings to host clusters.
– Enhanced support for throttles in terms of managing throttles per host, per host clusters, and for volumes.
– Introduced commands for CLI to manage changed volumes for Global and Metro Mirror relationships.
•Hardware change:
– Dense Drawer, which provides 5U rack-mounted high-density disk expansion enclosures that connect by using SAS to IBM SAN Volume Controller nodes. Each Dense Drawer can support up to 92 drives (model 2145-92F).
With V8.1, IBM has announced the following enhancements:
•Software change:
– Redesigned management GUI to conform to the look and feel of other strategic IBM products.
– Availability of Hot Spare Nodes in a cluster of 3 or more SVC nodes. In the event of an online node failure, the procedure to swap it by standby SVC node takes place.
– SVC now contains a software toolset that can establish a secured tunnel connection to the Remote Support Server that sits in an IBM internal network and collects health and troubleshooting information for support personnel. The associated Health Checker provides customized, rules-based configuration recommendations at no additional cost.
– Storing encryption keys to up to four external key servers that can standby with each other in case of a single key server failure. SVC supports co-existence of USB keys and external key servers.
– IBM Spectrum Virtualize in SVC can now address up to 256 GB of cache.
– Improved performance with dual-socket configuration. Before V8.1, even fully equipped SVC kept a second socket reserved for Real-time Compression (RtC) purposes even if this feature was not used. Starting with V8.1, SVC uses both sockets for non-RtC operations.
– Discontinued support of 2145-CF8 and CG8 models in V8.1. These models are supported up to V7.8 until their effective hardware End of Support date.
•Hardware changes
– Support of up to a maximum 16 x 10 GbE or 8 x 25 GbE Ethernet ports for increased iSCSI host logins. This feature reduces the number of rejected iSCSI logins from certain host platforms when all four allowed logins per host iqn are already in place.
– IBM SVC and Storwize Family delivers an additional option with a 900 GB 15K RPM 2.5-inch SAS drive.
1.4 Summary
The use of storage virtualization is the foundation for a flexible and reliable storage solution helps enterprises to better align business and IT organizations by optimizing the storage infrastructure and storage management to meet business demands.
IBM Spectrum Virtualize running on IBM SAN Volume Controller is a mature, ninth-generation virtualization solution that uses open standards and complies with the SNIA storage model. IBM SAN Volume Controller is an appliance-based, in-band block virtualization process in which intelligence (including advanced storage functions) is ported from individual storage devices to the storage network.
IBM Spectrum Virtualize can improve the usage of your storage resources, simplify storage management, and improve the availability of business applications.
1 Source: Storage Networking Industry Association.
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.