Special Section: equallogic iScSi peer Storage

including storage systems, RAID groups,                               the VM. This capability helps eliminate the         ...
Special Section: equallogic iScSi peer Storage

    iScSi and vMware infrastructure 3                 management, incre...
to be fully redundant, containing disks,
                                                           EqualLogic PS Series a...
Special Section: equallogic iScSi peer Storage

    automatic load balancing                                        gen...
enabling fast and flexible storage provi-
sioning, PS Series solutions can help dra-                                      ...
Upcoming SlideShare
Loading in...5

Flexible Virtualization with equallogic PS SerieS iScSi ...


Published on

Published in: Technology
  • Be the first to comment

  • Be the first to like this

No Downloads
Total Views
On Slideshare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

Flexible Virtualization with equallogic PS SerieS iScSi ...

  1. 1. Special Section: equallogic iScSi peer Storage Flexible Virtualization with equallogic PS SerieS iScSi Storage arrayS By Timothy Sherbak The advent of Internet SCSI (iSCSI) technology, combined with the latest in server virtualization software, can offer enhanced capabilities and benefits to enter- prises of all sizes. This article provides guidance on build- ing a virtualized infrastructure with EqualLogic® PS Series iSCSI storage arrays and VMware® virtualization software. I nternet SCSI (iSCSI) is a flexible and powerful stor- OS and its unique execution variables and associated age area network (SAN) protocol that can deliver data sets are maintained in these files. superior capabilities and benefits for enterprises Given sufficient memory and disk storage, a single of all sizes. In addition to providing enterprise-class industry-standard server such as a Dell™ PowerEdge™ data availability and performance, the iSCSI protocol server running virtualization software can support many enables breakthrough virtual storage designs that par- guest VMs. Applications and data are processed in allel the advanced designs of server virtualization tech- exactly the same way as in the physical environment, nologies such as VMware Infrastructure 3.1 using the same OS facilities for making system requests. Enterprises of all sizes are building flexible stor- The only difference is that the OS is now a virtualized age infrastructures using iSCSI and advanced virtu- guest running in a virtual system environment. alization technologies that let them allocate and shift VM files can be transparently migrated from one SAN resources dynamically in response to the server hardware platform to another, giving adminis- Related Categories: demands of their virtualized server environments. trators the ability to allocate and shift physical server EqualLogic This article describes a virtualized infrastructure that resources in response to changes in application work- Internet SCSI (iSCSI) applies storage and server virtualization technolo- loads. A scalable, highly resilient, flexible server envi- Storage gies to cost-effectively achieve a flexible, high per- ronment is created, helping increase resource utilization formance, dynamic IT infrastructure that is simple and IT flexibility and reduce operating costs. Virtualization to manage and scale. Storage virtualization is typically defined as a Visit DELL.COM/PowerSolutions for the complete category index. technology that allows discrete storage systems to Server virtualization operate as a single resource. In light of recent and Storage virtualization advancements in server virtualization, the concept of Server virtualization is one of the most significant storage virtualization is being further refined as a way technology trends in IT today. Managing server hard- to create an abstraction layer between the storage ware and software as separate infrastructure compo- hardware and logical data volumes. nents makes it possible to significantly increase data Given sufficient protocol support (as with iSCSI), center operating efficiencies. At a high level, server virtual storage products are now being designed that virtualization works by encapsulating guest operating allow data volumes to be located and striped across systems into a set of virtual machine (VM) files. The multiple (and diverse) physical storage resources, 1 For more information on iSCSI and virtualization, see “iSCSI: Changing the Economics of Storage; Part 2—Deploying iSCSI in Virtualized Data Centers,” by Matt Baker and Travis Vigil, in Dell Power Solutions, August 2007, DELL.COM/Downloads/Global/Power/ps3q07-20070401-Baker.pdf. 42 DELL POWER SOLUTIONS | February 2008 Reprinted from Dell Power Solutions, February 2008. Copyright © 2008 Dell Inc. All rights reserved.
  2. 2. including storage systems, RAID groups, the VM. This capability helps eliminate the fully redundant hot-pluggable compo- disk types, and controllers. This approach time-consuming necessity of copying VM nents, online hot-spare disks, environmen- not only helps increase performance and files, applications, and data from one ESX tal monitoring, and enterprise-class disk scalability, but also allows data volumes to Server system to another. drives with RAID protection. Furthermore, be transparently moved from one set of Networked storage (iSCSI or Fibre advanced availability features, including resources to another without disruption to Channel) is required to enable powerful fea- storage controller and I/O path failover, the operating systems and applications that tures within VirtualCenter, like VMware are required to help guarantee data access are utilizing the data. A scalable, highly resil- VMotion™ technology, which enables the even in the case of component failure. ient, flexible storage environment is formed, online migration of active VMs without Consideration should also be given to helping increase storage utilization rates interruption, and Distributed Resource evaluating the data protection software and significantly reduce operating costs. Scheduler, which monitors and automati- features enabled within the storage system. cally migrates VMs to help balance work- Storage consolidation within a SAN enables networked Storage with loads across resources within an ESX Server the consistent application of data protec- vMware infraStructure 3 cluster. VMware High Availability (VMware tion and disaster recovery, assuming the In a VMware Infrastructure 3 environment, HA), which also requires networked stor- basic SAN-based tools are available and multiple physical servers are networked into age, monitors the functioning of ESX Server enabled. Space-efficient, nonintrusive, a cooperating set of computing resources systems within the cluster, and in case of a snapshot-based point-in-time copies as well (see Figure 1). In lieu of a standard Microsoft® system failure, can re-host and restart VMs as efficient array-based replication tools Windows® or Linux® OS, each of the x86- affected by the failure on another ESX should be considered basic requirements of based standard servers run a copy of Server system within the cluster. the storage infrastructure for a virtualized VMware ESX Server, which provides the In addition to the requirement for net- environment. Multiple use cases exist for the abstraction layer between the physical serv- worked storage, a virtualized environment application of these tools in virtualized envi- ers and guest operating systems. Multiple heightens the need for high-performance, ronments, including disaster recovery, ESX Server systems are organized into a highly available, resilient storage to help online backup and quick recovery of VMs, cluster that can be controlled and coordi- meet the needs of the aggregated work- simple extraction of lost or corrupted data nated by VMware VirtualCenter, a central- loads. As critical applications, production files, rapid VM and data set provisioning ized management system. workloads, and data assets are consoli- using snapshots and clones for production Within the virtualized environment, the dated into relatively few resources, the and temporary use, and server-less tape- VM configuration parameters, OS, data need for high performance, nondisruptive and disk-based backup and recovery of the sets, and applications are encapsulated in scalability, and continuous availability of organization’s data assets. a set of files managed by the VMware the storage assets increases. Enterprise- A thorough evaluation of the storage Virtual Machine File System (VMFS). VMFS class storage designed for mission-critical platform should be made prior to deploy- is a cluster file system that coordinates deployments is a basic requirement when ment, reflecting not only on the robustness access across ESX Server systems, enabling building a virtualized IT infrastructure. and feature set, but also on the total cost of the migration of live VMs across systems In particular, purpose-built storage ownership over its useful life, with particular in the ESX Server cluster. architectures should be deployed that consideration of the required system growth The power of this architecture comes include mirrored memory write caches, in terms of both performance and capacity. from the ability to run virtually any guest OS, its applications, and its data without modifying the physical servers in the clus- Application Dell PowerEdge Application ter. As the need arises, additional server OS servers OS resources can be nondisruptively added to Application Application OS Application the ESX Server cluster, and current work- OS OS Application Switched Application Gigabit Ethernet loads are load balanced to take advantage OS OS iSCSI SAN of the newly available resources. Application Application Application OS Effective VMware virtualization OS OS deployments depend on shared access to Application Application OS OS storage—in other words, a SAN. SANs help ensure that each ESX Server system has ready access to the data sets of other VMs, enabling virtually immediate re-hosting of Figure 1. VMware ESX Server cluster with SAN storage Reprinted from Dell Power Solutions, February 2008. Copyright © 2008 Dell Inc. All rights reserved. DELL.COM/PowerSolutions 43
  3. 3. Special Section: equallogic iScSi peer Storage iScSi and vMware infrastructure 3 management, increase storage utilization, a World Wide Name (WWN), a physical With the advent of VMware Infrastructure provide virtually unlimited scalability in port address specifically assigned and 3, iSCSI initiators are integrated into the ESX capacity and performance, and enable encoded within the device hardware or Server kernel, allowing native access to online migration of data sets among con- firmware. WWNs are assigned for all devices iSCSI storage directly from ESX Server sys- trollers and storage tiers. By consolidating within the SAN, including each physical port tems. Both hardware initiators and software network-based storage into a simple, flex- within each host server and each physical initiators are available. Network interface ible, consolidated pool of storage designed port within the storage controllers. card teaming (multiplexing a single logical to grow in capacity and performance (as Furthermore, data paths between the hosts connection across multiple interfaces) is in an EqualLogic storage environment), and the storage array are statically set when supported with the software initiator. The storage virtualization can help provide an the host is added to the SAN. hardware initiator includes support for multi- exceptional virtualization deployment. Herein lies the inflexibility of Fibre pathing as well as boot from SAN. Historically, the adoption of storage vir- Channel–based SANs. The topology of the iSCSI support broadens the potential for tualization technologies in SAN environ- SAN is intrinsically hard coded into the envi- full-fledged VMware virtualization deploy- ments has been limited to enterprise data ronment, making changes within the infra- ments in small and medium businesses by centers needing a tool for online data structure burdensome, time-consuming, helping remove the need for a costly and migration between heterogeneous storage and error prone. Expansion of resources in complex Fibre Channel–based SAN deploy- environments. This add-on design allows for the host, fabric, or storage layer propagates ment. The acquisition, implementation, and heterogeneous storage and data move- changes throughout the infrastructure, operational costs of an iSCSI-based SAN ment, but does not simplify management causing intrusive downtime to the applica- are intrinsically lower than those of a Fibre of the configuration. Most of these designs tions and infrastructure. Channel–based SAN, helping remove an have been implemented as external appli- In contrast, TCP/IP networks support economic barrier to the adoption of server ances sitting within the SAN. These devices virtual addressing and dynamic routing, virtualization technologies. Advanced iSCSI- generally add complexity to an already whereby paths through the network are not based system architectures can help dra- complex environment—introducing addi- statically defined. With Dynamic Host matically change enterprise expectations tional points of management within the Configuration Protocol (DHCP), addresses for large deployments as well. A truly flex- SAN, masking value-added features of the can be dynamically assigned, or through ible iSCSI-based virtualized environment attached arrays, and limiting the perfor- address proxying, physical addresses can can help not only reduce costs, but also mance and scalability of the consolidated be virtualized, making it possible for devices simplify deployment, provide comprehen- SAN storage pool to a single gateway host- in the network to transparently act on sive storage management and data protec- ing the back-end storage. behalf of other devices on the network. If tion functionality, and enable seamless VM Advanced storage virtualization tech- the IP address of the resource being mobility between hosts. nologies, like those available with PS Series accessed is known from within the network, The addition of iSCSI protocol interfaces arrays, offer the ability to virtualize storage the entity (such as a SAN initiator) request- to classic storage system designs does not at the storage device level. This approach ing access can find the resource dynami- fully exploit the potential of the iSCSI pro- helps achieve scalable performance and cally without prior knowledge of the paths tocol and, in turn, does not adequately solve capacity and reduce overall management or the resource’s physical address. the storage management challenges faced by aggregating multiple controllers into a IP address proxying helps make stor- today by IT managers. Even with the intro- cooperating set of resources—that is, by age access fully virtualized across multiple duction of iSCSI, storage administrators virtualizing volumes not only across disks EqualLogic PS Series storage controllers. using classic storage systems typically within a controller but also across storage In an EqualLogic group, for example, each require a rarified level of knowledge for systems in a SAN. member array has 3 active physical configuring and tuning storage arrays, RAID EqualLogic PS Series iSCSI storage Gigabit Ethernet ports; a four-member geometries and data layout considerations, arrays are an excellent example of such a group has a total of 12 active Gigabit application workload analysis, forced data “scale-out” architectural design. Despite Ethernet ports, providing 12 Gbps of peak migration, and complex system upgrades. their powerful simplicity, such designs are bandwidth. Each of these Ethernet ports virtually nonexistent in the Fibre Channel– is assigned an IP address. The members advanced virtualization based SAN world, primarily because of within the group are aware of and coor- with the equallogic pS SerieS architectural constraints inherent in Fibre dinate use of these multiple IP addresses Storage virtualization, a key feature of Channel network deployments. and the underlying physical resources. EqualLogic PS Series arrays, can help For example, devices in a Fibre Channel– External access to the group by all hosts simplify provisioning and ongoing based SAN are attached to the network via within the SAN is exclusively addressed 44 DELL POWER SOLUTIONS | February 2008 Reprinted from Dell Power Solutions, February 2008. Copyright © 2008 Dell Inc. All rights reserved.
  4. 4. to be fully redundant, containing disks, EqualLogic PS Series arrays multiple high-performance network inter- Volume faces, redundant controllers with mir- Volume rored battery-backed caches, and other advanced features. The disks can be Volume Volume automatically protected with RAID Switched Volume Volume (RAID-5, RAID-10, or RAID-50) and hot Gigabit Ethernet spares. Multiple models of arrays exist, iSCSI SAN Volume Volume leveraging high-performance 10,000 rpm Volume and 15,000 rpm Serial Attached SCSI Volume (SAS) drives or cost-effective, high- Volume density Serial ATA (SATA) II drives. Volume A PS Series storage group comprises a single array or multiple arrays. A group is a virtualized resource, appearing to ESX Figure 2. EqualLogic virtualized iSCSI-based SAN Server systems as a single entity that offers network storage access to a single large through the group IP address, a unique IP and disks to help extend performance and pool of storage—a SAN composed of a address that transcends all of the underly- capacity. Multiple tiers of storage can be single virtually scalable, high-performance ing member IP addresses. The only IP seamlessly integrated in the utility, provid- storage system. Unlike a conventional SAN, address known by the attached hosts in ing for automatic placement of data within in which multiple controllers appear as the SAN infrastructure is the group IP a volume to the appropriate tier based on independently managed islands of storage, address. I/Os may be serviced by any the volume’s access patterns. each group member in an EqualLogic SAN physical Gigabit Ethernet port. A flexible storage architecture is par- can cooperate with other members to Shielding host systems within the SAN ticularly appealing in a virtualized server automate resource provisioning and per- from being aware of the physical topology environment because the scale-out archi- formance optimization. of the storage serving its data is vitally tectures of both server and storage infra- ESX Server administrators can create important, particularly in a virtualized server structures complement each other, data volumes from within the EqualLogic environment. By virtualizing the physical providing an end-to-end virtualized infra- group storage pool. The group exports I/O ports, a flexible storage utility is created. structure. By deploying both virtualized its volumes as iSCSI targets protected Pooling storage assets within the SAN is servers and virtualized storage together, with security measures that include now possible, enabling a rich set of features IT managers achieve an infrastructure that authentication and authorization. Upon to help maximize storage utilization and is particularly adaptable to changing busi- connection, ESX Server systems work effectiveness, extending beyond the limita- ness requirements by providing nondis- with EqualLogic volumes as VMFS data tions of a single physical end-to-end con- ruptive workload migration and balancing stores, which contain multiple VMs and nection to a single storage controller. across all physical resources (both stor- virtual disks. A variation on this is a raw Data is accessed by the host without age and server resources) as well as online device map in which the SAN-resident intimate knowledge of where in the stor- resource expansion that can immediately volume is mapped directly to a corre- age utility the data is stored, providing the apply to preexisting server and storage sponding virtual disk within the ESX opportunity for data volumes to span workloads without intervention and Server environment, circumventing the controller boundaries and exploit the tuning. For more information on the VMFS data store. A third alternative is to combined resources of multiple control- advantages of this type of environment, employ the native iSCSI software initiator lers. Controllers coordinate among them- see the “Integrating server and storage of the individual guest OS to connect selves to help balance workloads and virtualization” sidebar in this article. directly to the SAN-resident volume. This optimize storage resource utilization. last option allows administrators to use Changes in storage infrastructure can equallogic virtualized iScSi-based Sans all array-based functionality and allows be achieved seamlessly, without affecting A virtualized iSCSI-based SAN comple- seamless integration with high-level host connectivity or data access. The ments a VMware ESX Server farm. The applications, particularly backup applica- iSCSI-based infrastructure can be scaled foundation of an EqualLogic virtualized tions that use Microsoft Volume Shadow nondisruptively, automatically applying iSCSI-based SAN is the PS Series storage Copy Service (VSS) or Virtual Disk Service the resources of the additional controllers array (see Figure 2). Each array is designed (VDS) as an underlying protocol. Reprinted from Dell Power Solutions, February 2008. Copyright © 2008 Dell Inc. All rights reserved. DELL.COM/PowerSolutions 45
  5. 5. Special Section: equallogic iScSi peer Storage automatic load balancing generated by the ESX Server farm. Data members “learn” configuration and perfor- Volumes are distributed among the PS and network I/O to the group can be mance information from the group—with Series group’s member arrays, with data automatically load balanced across the no manual intervention. Data and client placement and access continually group members’ resources. connection load balancing can occur auto- adjusted for optimal performance as A key feature of PS Series arrays that matically as the group scales. I/O activity resources are added or workloads change. sets them apart from traditional storage is monitored, and data and network con- When an array is added as a group is their dynamic scalability and automated nections are adjusted as needed. member, its disk space is added to the configuration, enabling administrators to group’s storage pool. Volumes can be flexibly meet changing needs while main- nondisruptive scalability automatically re-striped and distributed taining availability. As capacity and per- The scalability model allows for auto- across the members of the storage pool. formance requirements increase, a group mated, online expansion across storage Controller resources are also dynamically can be scaled linearly in both capacity dimensions, and the EqualLogic PS Series adjusted based on the workloads being and performance—all while online. New architecture helps nearly eliminate down- time caused by expanding or managing a storage system. Because capacity can be integrating SerVer and Storage added so easily, IT managers need to buy Virtualization only the storage necessary for today’s applications, helping ease budget con- Combining server virtualization with the storage virtualization of EqualLogic PS Series iSCSI arrays straints caused by excessive purchases. can enable organizations to create powerful, highly flexible virtualized environments, offering a variety Additional advantages of the virtual- of advantages: ized SAN environment include the trans- parent application of storage tiers. ■■ Aggregation of virtualized assets on consolidated hardware: Operational procedures and EqualLogic storage groups can use mul- best practices can be standardized and consistently applied to both storage and server assets tiple storage tiers contained within the (both physical and virtual), helping provide a resilient infrastructure with high resource utilization, storage pool and automatically optimize high levels of service, and enhanced protection of information assets. data placement based on workload. ■■ Simple, centralized management: Increased management efficiencies can be achieved by Alternatively, storage tiers may be seg- centralizing management through intuitive, graphical management tools accessible from virtually mented into independent resource pools to help guarantee specific resources to anywhere on the network, providing a comprehensive view to provision, monitor, and manage specific workloads, concurrently maintain- the entire virtualized infrastructure. ing the flexibility to migrate volumes ■■ Flexible and quick deployment of virtualized resources: Organizations can quickly adapt to online from one tier to another, transpar- changing and growing business needs by reducing the time to provision and deploy new applica- ently to the ESX Server hosts. tions through quick provisioning methods available for both servers and storage. ■■ Online, nondisruptive resource reallocation and expansion: As workflows and business quick, intelligent provisioning priorities change, both storage and server resources can be simply reallocated online, with no Within the EqualLogic environment, stor- disruption to operations. In addition, physical resources can be easily expanded online without age is quickly provisioned simply and easily downtime. Workloads can be automatically rebalanced across these newly available resources with the click of a button. The key func- without disruption to applications. tions needed to configure, manage, and scale storage are automated, helping cut ■■ Common IP network–based infrastructure: The IT environment is simplified by basing opera- administration tasks for volume and capac- tions on IP networking, including the interconnect for client access, inter-server communication, ity growth. Given the intelligence built into storage access, and off-site data replication. The organization’s inherent IP networking expertise the PS Series software, decisions with is leveraged, helping reduce training and ongoing management costs. respect to RAID type and data layout can ■■ Enterprise-class resiliency: With redundancy built into the physical server, network, and storage be made automatically at the time of pro- architecture, as well as component failure detection and failover software implemented within visioning, and optimized as workload pat- each layer of the infrastructure, overall reliability, availability, and service levels are enhanced. terns for the newly provisioned storage ■■ Advanced data management and disaster recovery: A rich set of server and SAN-based evolve. As new resources become avail- data protection tools help ensure the organization’s critical assets are protected and immediately able, the resources can be automatically recoverable at the local or a remote site. applied where and when they are needed. By helping eliminate complex tasks and 46 DELL POWER SOLUTIONS | February 2008 Reprinted from Dell Power Solutions, February 2008. Copyright © 2008 Dell Inc. All rights reserved.
  6. 6. enabling fast and flexible storage provi- sioning, PS Series solutions can help dra- EqualLogic PS Series arrays Dell PowerEdge servers matically reduce acquisition and ongoing Volume Application operational costs and make enterprise- OS Volume class shared block storage practical for Application OS Volume organizations of all sizes, from small busi- Application Volume nesses to large enterprises. OS Volume Volume Application Switched Thin provisioning extends existing OS Gigabit Ethernet Volume iSCSI SAN EqualLogic provisioning features, helping Application OS Volume Volume make the “buy-as-you-grow” storage Application model of the modular PS Series seamless OS Volume Application for servers and applications. Thin provi- OS Physical Volume infrastructure sioning is an important advanced feature that enables the automatic addition of Virtualized assets physical capacity on demand up to preset limits. With advanced thin provisioning, buy-as-you-grow storage management Figure 3. Virtualized servers combined with virtualized storage and virtualization are made seamless for servers and applications. When a volume manually map application data to specific enterprise data centers, a virtualized SAN is created, it can be sized for the long- physical devices and controllers. utilizing PS Series arrays offers a cost- term needs of the application without ini- A primary advantage of this auto- effective, comprehensive solution to help tially allocating the full amount of physical mated virtual storage system is that stor- reduce both management complexity and storage. Instead, as the application needs age administration remains stable as the total cost of ownership. additional storage, capacity is allocated group is expanded. All administration is to the volume from a free pool. performed at the group level, through the Timothy Sherbak is senior manager of vir- EqualLogic thin-provisioning capabili- EqualLogic PS Group Manager, an intui- tualization solutions marketing for the Dell ties help provide extensive flexibility and tive, single-pane administrative console. EqualLogic product family. He has over 20 user safety controls. These include the Because the group is managed as a single years of engineering, sales, and marketing ability to turn thin provisioning on and off logical system, the ongoing operational experience in computer systems and storage as needed for any volume, allowing users costs of storage management can remain networking technologies. Timothy has a B.S. to cost-effectively test the most suitable fixed even as storage grows. in Systems and Information Science from applications and volumes for thin provi- Syracuse University and an M.B.A. from the sioning, with the knowledge that they can SiMple, coSt-effective, Massachusetts Institute of Technology return to “normal” provisioning online. dynaMic it environMent Sloan School of Management. EqualLogic implementation of thin provi- Virtualized iSCSI-based SANs, such as sioning also provides enhanced alerts and those enabled by EqualLogic PS Series controls—with proactive, user-defined arrays, are changing enterprise expecta- threshold alarms and controls, administra- tions of how simple a storage infrastruc- tors can depend on automatic space allo- ture can be to deploy, manage, and grow. cation without worrying about reaching As Figure 3 illustrates, complementary allocation limits or unexpected depletion server and storage virtualization built on of physical storage. iSCSI-based SANs create a fully abstracted quick linkS pool of physical resources—one that can automated management provide higher performance, scalability, EqualLogic: The PS Series architecture is designed to ease of use, and flexibility compared with www.equallogic.com simplify storage management in several classic storage array architectures. EqualLogic and VMware: ways. RAID configuration and hot sparing is EqualLogic PS Series iSCSI arrays are www.equallogic.com/vmware automated, and dynamic storage and net- designed to reduce technical complexity EqualLogic PS Series work I/O load balancing can occur automati- and cost barriers imposed by classic SAN demonstrations: www.equallogic.com/demo cally as resources and performance metrics architectures without compromising change. No longer must administrators performance, scalability, or resiliency. In Reprinted from Dell Power Solutions, February 2008. Copyright © 2008 Dell Inc. All rights reserved. DELL.COM/PowerSolutions 47