Technology Overview and Use Cases
Technology Overview and Use Cases
Abstract
The HP StorageWorks SAN virtualization Services Platform or SVSP is a network-based storage virtualization solution that enables enterprises to manage, pool, and share storage resources to improve efficiency, simplify operations, and lower total cost of ownership (TCO). This paper provides a general overview of virtualization technology and explains in more detail the value of storage virtualization in modern IT infrastructures. It then describes the HP SVSP architecture, its benefits, and the various data management services that SVSP provides. Finally, this paper presents a number of common use cases and describes more specifically the benefits of the SVSP in those scenarios.
Introduction
In todays challenging business environment, IT managers are seeking infrastructure solutions that extend the value of current and future IT investments, simplify operations, and increase the overall efficiency. As IT looks for transformation to bring about such a vision, virtualization, and especially server virtualization, has emerged as one of the most significant shifts in IT. Server virtualization adoption is high and it is being used extensively in both mission critical and business critical applications. It provides well recognized benefits like increased server utilization, the ability to easily provision compute resources, and improved application availability. However, it brings a series of storage challenges that could undermine and even nullify the benefits of server virtualization. Virtual environments need virtualized storage to enable and actually accelerate the benefits of server virtualization. Network-based storage virtualization in particular, provides a flexible, scalable implementation that enables firms to optimize capacity utilization and streamline operations for greater productivity. This paper reviews the SVSP, newest network-based virtualization solution by HP, and explores in more details the specific ways in which storage virtualization can help IT to drive better business outcomes.
Virtualization Overview
What is virtualization?
Virtualization is the aggregation of physical resources into a unified structure (pool) and the presentation of those resources as capabilities that can be consumed by applications or other types of clients. The term virtualization has really become an umbrella that encompasses a vast group of encapsulation and management technologies as well as implementation methods. These approaches create pools of sharable resources that enhance utilization and, ultimately, can automatically allocate the resources to match supply to demand.
Storage virtualization can create logical views of storage that are distinct from their physical components, thereby insulating application access and utilization of storage from infrastructure configuration and management. The term storage virtualization is used to indicate the operation of dividing the available storage space into virtual volumes without regard to the physical layout or topology of the actual storage elements (for example, disk drives, RAID subsystems, and the like). Typically, virtual volumes are presented to Operating Systems as an abstraction of physical disks and are used by these OS as if they were such disk drives. The main difference between a virtual volume and physical volume is that a virtual volume can be created, expanded, deleted, moved, and selectively presented, independent of the underlying storage subsystems. The independence of virtual volumes from physical storage is essential to support the dynamic mobility inherent in virtual machines. It is also essential for maximizing storage utilization (with features like Thin Provisioning), and to enable the migration and replacement of obsolete storage subsystems without causing data access interruptions. In addition, virtualization of storage enables the usage of software applications instead of human administrators to perform volume allocations. Finally, data management is significantly enhanced through storage applications such as Snapshots, Remote Mirroring, and Server-Free Backup. To summarize, the promise of SANs can be fully realized with storage virtualization.
Speedy delivery of new applications and business services Put unused resources and capacity to work Increase focus on business priorities Reduce IT costs and risk
Typical block virtualization benefits (1) Improved asset utilization by 300% Shrink back-up windows by up to 80% Manage up to 3x more storage/administrator
(1)
Business Value of Storage Virtualization: Scaling the storage solution, leveraging the storage investment, Richard Villars & Randy Perry, IDC, Feb. 2009
network infrastructure
Symmetric (In-the-Data-Path) virtualization The general architecture of the symmetric approach (sometimes referred to as In-the-Data-Path) is shown in the following figure. This approach calls for the appliance (a computing platform and associated memory) to be installed between the servers and the storage resources on the SAN, at least in terms of data flow. A key drawback of the symmetric approach to virtualization is that it creates a SAN bottleneck (and thus limits the SAN performance and scalability) and significantly complicates the design of large-scale highly available configurations.
The symmetric virtualization concept creates a bottleneck because all data from all applications must pass through a virtualization appliance to be processed, and thus forces the appliance to sustain the throughput of the entire SAN, impacting performance. Some symmetric designs are attempting to circumvent this performance problem by adding a cache facility to the appliance. However, the use of caching within the fabric plays against the high availability and scalability qualities of the SAN. In order to scale, complex mechanisms must be put in place to insure cache coherence. Experience with RAID controllers has shown that this can be both complicated and expensive. Split path virtualization This method uses a combination of appliances and network agents to create and manage virtual volumes while enabling direct data transfer between server and storage subsystems for optimal performance. By having multiple storage subsystems working in parallel with multiple servers, total performance can be increased up to the maximum of the FC fabric bandwidth. In the split path approach metadata handling is done separately from the data path. The appliance serves as a Metadata Center that sees the physical storage and allocates virtual volume mapping.
The agent retrieves the volume configuration from the appliance and presents virtual volumes to the operating system as if they were disk drives. When the operating system sends an I/O to the virtual volume, the Agent intercepts the I/O, translates the volumes logical address to the physical address, and sends the I/Os directly to the storage devices. The HP SVSP is an example of this architecture, where the Data Path Modules (or DPMs) act as agent and the Virtualization Server managers (or VSMs) act as the appliance. This architecture maintains the flexibility of symmetric virtualization without incurring the degradation of performance or the high cost of hardware. The appliance can be a small and inexpensive unit.
High Availability (HA) SAN configurations can be implemented through simple redundancy of the appliance. Very high degrees of scalability can be achieved through loosely connected storage domains and similar designs, benefiting from the key advantage of the split path concept, so that there are neither pending I/Os stored in the appliance nor is there a need for clustering, cache coherency algorithms, or other complex mechanisms, which limit the practical expansion of real-life enterprise SANs.
HP SVSP at a glance
The HP SVSP is targeted at enterprise and mid-range firms that are experiencing significant storage growth, may be deploying server virtualization, and need to improve productivity. The SVSP helps customers improve efficiency, simplify operations, and lower TCO, by enabling the pooling and sharing of storage resources and providing a range of storage services for those resources. The SVSP also supports heterogeneous arrays to facilitate pooling of storage across HP and non-HP arrays. The SVSPs comprehensive set of data service includes: Centralized management and pooling of storage resources Consolidation of HP and non-HP arrays Centralized volume management High performance, highly scalable split path architecture Comprehensive set of storage data services Non-disruptive data migration copy services (clones and snapshots) Sync (local) and Async (local and remote) mirroring Improved asset utilization and decreased costs with thin provisioning Provision thin volumes and grow to larger ones over time Thin provisioning for any vendors hardware No rip-and-replace approaches are required
Figure #5 illustrates a typical SVSP SAN deployment. The storage layer includes arrays from different vendors (HP, EMC, SUN, and IBM) and a Fibre Channel SAN (the storage network itself). The server layer has multiple servers from different vendors running different operating systems: Windows, Linux, HP-UX, and the like. The figure also illustrates the detail of the split path architecture for the SVSP as well as the presence of the typical data services: snapshots, snapclones, data migration, thin provisioning, and remote mirroring. Storage in an SVSP system starts with LUNs from each of the storage arraysthe LUNS are presented to the SVSP system, which will manage them for day-to-day operations. The virtualization work is performed by the DPMs and VSMs in the fabric (at least two of each for redundancy). The DPM is the element in the architecture that does real-time parsing of FC frames by examining packets and making immediate translation decisions. The DPM gets its virtual-to-physical storage mappings from the VSM software that performs data management operations (for example, backup, migration). In this way, the SVSPs innovative design separates data flow (through DPM) from management processes (handled by VSM), so that the processing power needed to handle data path workloads can scale independently from the resources used for management and data movement. All management or control path operations take place between the VSM and DPM without disrupting the data path between servers and storage arrays.
The SVSP allows for the creation of storage pools according to business needs and required Quality of Service. (For example, in figure 5, the storage pool marked green can be used to store data with high availability requirements.) From those storage pools, the platform can create virtual disks which are assigned to the different servers. The SVSP allows better utilization of the storage by assigning free storage from the storage pools. Additionally, users can choose to thin provision volumes for greater asset utilization. Storage pool expansion becomes routine and simple. Finally, the SVSP also delivers a unified set of data services that provides the necessary tools to address a series of important use casesexplained in the next sectionthat drive significant value by helping customers improve efficiency, simplify operations, and increase productivity. The remainder of this paper will review these use cases.
Problems provisioning heterogeneous HA SANs Problems testing new applications Not enough time to backup or the need for rapid recovery Migrating data between heterogeneous storage Struggling to create a DR Plan
Clones and Snapshots Non-Disruptive Data Migration with SVSP Volume Manager
Use Cases
Consolidation and Centralized Management
Improving IT economics with storage virtualization Many customers today have storage arrays from different vendors or at least different generations of arrays from the same vendor. Except in very rare instances, the customer is forced to recall the capabilities of each type of storage and use a different set of management tools or licenses from each vendor, including data services options, in order to manage each array separately. As the number of servers (virtual and physical) and storage devices grows, hundreds or thousands of volumes must be individually monitored and managed by storage administratorsa daunting task to say the least. Complexity, cost, and capacity utilization rates are all negatively impacted in these traditional deployments, particularly when server virtualization is introduced.
10
Customer Problem
Management complexity of multiple Poor storage capacity utilization Application downtime due to
Solution
Aggregate capacity through
network-based virtualization with the SVSP, to form centrally managed pools of virtual storage unused space, maximize application uptime by expanding capacity on a just-in-time basis,
Time consuming provisioning of High cost of a single tier storage Lack of ability to respond faster to
Operational impact
Prevents server downtime due to
multiple data services with a consistent set of tools data among the tiers to optimize storage costs
out-of-space conditions
virtualization requirements
provisioning for all capacity under management irrespective of the array type
using one management console for all storage devices striping across RAID arrays
Financial Impact
Reduces storage costs
Reclaims unused storage Greater storage utilization Lower TCO, CAPEX, and OPEX
Consolidation and management of multiple arrays from various vendors is made easy using the SVSP. It allows the storage administrator to aggregate capacity through virtualization into centrally managed pools. The SVSP facilitates management of heterogeneous arrays, which can be used to create storage pools and create front-end (host visible) volumes. The HP SVSP also deploys role based management, which makes day-to-day administrative jobs easier. Storage consolidation and centralized storage management with the HP SVSP Aggregates all available capacity into centrally managed pools Improves performance by striping LUNs across RAID arrays Allocates appropriately-sized LUNs to each server Provides thin provisioning to increase storage utilization Maintains unused capacity in pool for future requirements Dynamically and granularly allocates additional capacity to any server on a just-in-time basis Strong multi-vendor support
11
12
Solution
HP SVSP on-line Data migration
according to its current value, as part of an ILM strategy and data from older, slower, or more expensive storage to newer, faster, or less expensive devices devices by proactively reassigning applications to different storage devices to achieve better performance and balance workloads copy before the migration process begins
capabilities enable any-to-any (heterogeneous) data migration while production applications remain online. It also protects data by leaving the original data intact for further use.
upgrades that support strategic goals and objectives of revenue generating applications
Operational Impact
Avoids the costs and disruption of
lost access to applications and data Migrates data to any devices at any location while production applications remain online Enhances data protection by using HP SVSP snapshot to create instant, read/write snapshot copies of migrating volumes
across all servers and storage devices Management (ILM) process, to maximize storage utilization
Financial Impact
Maximizes availability of
storage only for data that has high value. Move data when its value changes to avoid unnecessary expenses
The HP SVSP has a non-disruptive Data Migration capability which allows data to be migrated efficiently from one array to another, allowing applications to be online and fully available during the migration without server impact. This SVSP data service enables seamless movement of data among the storage tiers, making it easy to migrate data to cost-appropriate, performance-appropriate, and availability-appropriate storage pools that make sense to the business and optimize storage costs. The SVSPs data migration feature also ensures that the original LUNs remain intact for fail-safe roll back purposes. Using the import in place feature, existing storage can be quickly moved into the pools and presented through SVSP. These virtual disks can be used anytime and all the advanced features can be applied on them as a normal virtual disk.
13
Information Lifecycle Management (ILM) ILM is a process of treating your data according to its value. As part of this concept, data is placed on the appropriate storage device to match the performance, capacity, and availability requirements of that data. As business needs and data value change over time, an online volume migration tool is mandatory for ILM strategies. Moreover, data migration may be required from time to time. Refresh older storage to improve performance and lower costs Seamlessly transitioning production applications from older, slower, and more expensive-to-maintain storage to newer devices can result in significantly better performance at a lower cost. A migration data services is also essential to efficiently carry this type of IT operations
Non-disruptive, any-to-any data migration with HP SVSP data migration Any-to-any data movement, from any device, to any device, to any location Migrates data while production applications remain online. No change or activity is required by the host Integrates with HP SVSP Snapshot to create instant snapshots of migrating volumes for enhanced data production Original volume remains intact for backup purposes SAN-wide, centralized management of all migration activities Supports all major operating systems and storage devices
14
Solution
Use HP SVSP Business Copy solutions
and delivery software to improve business operations, maintain competitiveness, and generate new sources of revenue
maintains competitive advantages by rolling out new features faster by improving software quality
of current production data but without disrupting normal operations bugs and errors prior to deployment
Operational Impact
Prevents system disruptions
Creates copies and allows testing to occur without disrupting normal operations Enables more thorough testing to eliminate software bugs and errors prior to deployment
modify applications
Financial Impact
Reduces costs by eliminating
A lengthy test cycle extends new application development, which increases the development costs, leading to delayed revenue and loss of competitive market position. Lack of time and resources to thoroughly test more data permutations also affects the quality of any product. The SVSP allows faster deployment of new solutions. With the SVSP Business Copy services, IT personnel can instantaneously present a snapshot to a virtual or physical server, to start testing on copies of production data without disrupting production systems. Furthermore, the capabilities support the creation of snapshots that will enable you to keep copies of every stage in a multi-stage testing process. If something goes wrong in any stage, you can easily roll back to the previous stage without the need for repeating all stages. As a result, productivity will be improved by maximizing development time. A flexible, Adaptive Infrastructure enables faster responses to changing business conditions.
15
Accelerating application time-to-market Streamlining the application testing process can significantly accelerate the delivery of all types of applications. Figure 10, depicts application time to market. With HP SVSP, the copy and staging process is reduced to minutes for each test run, resulting in faster testing and faster time-to-market.
Before
After
Application testing with HP SVSP Business Copy Creates instant read/write copies of production data without taking applications offline Allows any-to-any data movement between production and test environments Allows multiple tests on multiple copies to occur in parallel Changes to test data do not affect production data In time, will support all major operating systems and storage devices
16
17
Customer Problem
Rapidly recover data in
Solution Use HP SVSP Business Copy feature to create frequent, instant, read/write disk-based, low capacity snapshots of production data. When necessary, recover from Point-in-Time (PIT) snapshot.
Recover a missing or corrupted file
minutes rather than hours (reduce RTO to minutes) disaster (reduce RPO to minutes) capacity growth (capacities are growing at 50% + per year) servers and storage
performance of key business applications enabling rapid resumption of operations after unexpected events
by mounting a recovery server to a view on a PIT and copy missing/corrupted file back rolling-back a snapshot
Operational Impact
Reduces Recovery Point Objective
satisfaction
Financial Impact
Maximizes availability of revenue
The SVSP Business Copy feature minimizes downtime by enabling rapid restore of volumes and files. Without the SVSP, data recovery can take hours and data loss is possible. Instead the SVSP provides Snapshot and Snapclone features, which can be used to create rapid recovery facilities. Copies of production data can be captured using snapshot or snapclone and can be restored easily. Rapid restore with SVSP business copy Create frequent, instant, low capacity snapshots on any storage array In case of a failure, mount any prior snapshot and examine its content To recover a file, simply copy the file from the snapshot to the production volume To recover the entire volume, mount the production server to snapshot This entire process can be completed within minutes
18
19
Clustered Solutions A common Microsoft Cluster Server configuration involves a SAN environment where both servers (see figure #15) access a single image on a storage subsystem. From a DR perspective, the standard configuration is lacking in several regards, including: No protection against the failure of the storage subsystem No protection against the failure of the site No failover/failback capability between sites
20
Customer Problem
Instantaneous data recovery in case
Solution
Use the Synchronous mirroring
of disaster
by mirroring data from any device, to any device, at any location mirrored volumes at any location by creating instant, read/write snapshots mirroring processes across multiple devices recovery site
capabilities in the SVSP Volume Manager to protect your data when you need 100% data protection. Centrally manage data mirroring activities across the enterprise and create instant read/write snapshots of mirrored volumes. Synchronous mirroring can be achieved between sites for mid-range distances. remote copy services of production data to ensure high availability. Restore entire applications in minutes with HP SVSP Business Copy. Use snapshot enhanced instant roll-back for improved data integrity
Operational Impact
Protects against operational downtime
Allows rapid failover and access to copies of data in secondary locations Allows systems at any location to be taken down for maintenance while applications remain online
HP SVSP Snapshot and Snapclone to create instant, read/write snapshots of mirrored volumes for use in zero-downtime backup and online restores mirroring of data for all servers and storage devices points in time, according to users definition and fail-back
Easy disaster recovery site testing Enhances and simplifies the mirroring
Financial Impact
Maximizes uptime of
21
HP SVSP for Disaster Recovery The synchronous mirroring features in the SVSP Volume Manager or the remote Async Mirroring capabilities in the SVSP Continuous Access option can facilitate instantaneous service resumption after storage, site, or regional disaster. The HP SVSP Business Copy feature, which will keep the system online and fully available during backup without impact, allows restoring the entire application in minutes. Together these SVSP data services help firms improve productivity through maximum uptime. Centralized management of recovery operations reduces administrative tasks and minimizes errors while lowering management costs.
22
SAN-based HP SVSP SolutionCampus deployments When two sites share a SAN, HP SVSP Volume Manager can be used to synchronously mirror data between the sites. In this configuration, there is one VSM and (at least) one DPM at each site, such that both sites are part of the same SVSP domain and have an identical view of the virtualized storage. Application servers run at both sites (in either clustered or primary/backup configurations), and each site has its own storage subsystem. With SVSP handling the synchronous mirroring across the two storage systems, the two sites do not need to have identical types or brands of storage. This solution provides business continuity after server, storage, or site failures, and it is usable up to mid-range distances. A common application is a cross-campus deployment within a metropolitan region. (Please consult the HP SAN design guide at www.hp.com/go/SANDesignGuide for detailed information about supported distance limits.)
Figure 17: Clustered application server configuration for dual site domain HP SVSP Solution
23
Low Cost Long Distance DR Solution - Snapshot-enhanced, any-to-any Data Mirroring In this scenario (see figure 18), servers are present across two different sites or domains which can be further apart than those in the SAN-based solution above. Like the SAN-based solution, each site has its own application servers and storage, but in this case each site is also a complete SVSP domain (that is each site has its own pair of VSMs and DPMs). The asynchronous mirroring capabilities of SVSP are used to replicate data across the domains in a bandwidth-efficient way, across existing IP connections. This solution provides disaster recovery after server, storage, or site failures across large geographic areas.
Data is asynchronously mirrored between the sites, using existing IP infrastructure. (Note the contrast to alternative solutions which often require dedicated and costly high-bandwidth links.) This solution provides: An affordable, vendor-independent solution that enables rapid recovery from both planned and unplanned downtime Mirroring from any device to any device, to any location Integrates with HP SVSP Business Copy feature to create instant, R/W snapshots of mirrored volumes Local and remote mirroring (LAN, MAN, and WAN) Optimized for high performance over limited bandwidth connections SAN-wide, centralized management of all mirroring processes Supports all major operating systems and storage devices Facilitates data integrity at the disaster recovery site Unique procedure for test disaster recovery site with easy and quick fail-back to the original site
24
Solution
Remote Backup Remote Snapshots Local MirrorSync SAN-wide Synchronous mirroring
Regional Disaster
Operational impact
Reduces disaster recovery efforts Enhances and simplifies the mirroring
Financial impact
Maximizes availability of
25
Applications must function 24x7 Add storage/ servers on the fly No unscheduled outages No scheduled outages Data recovery must be immediate Remote disaster recovery
26
Conclusion
HP SVSP is a comprehensive infrastructure solution that helps IT managers extend the value of current and future IT investments, simplify operations, and increase the overall efficiency of IT. The SVSP aggregates capacity of heterogeneous arrays and creates pools of virtual storage that can be easily provisioned to virtual and physical servers. The solution suite also provides a comprehensive set of virtualization enabled data services that include: volume management, copy services (snapshot and snapclones), non-disruptive data migration, sync/async mirroring, and thin provisioning. These data services improve storage utilization while simplifying storage management. Furthermore, they also enable advanced replication and management services, giving users virtually limitless opportunities to utilize up-to-date copies of production data for activities such as online migration, consolidation, rapid application recovery, zero window backup, and disaster recovery. This is a powerful tool kit from which solutions are built to solve the most challenging data management problems that IT managers face today. As a result, the SVSP represents a solid infrastructure solution that gives enterprises the ability to improve productivity, simplify operations, and achieve lower total cost of ownership for their SAN environments.
27