Consolidating file servers involves decisions regarding virtualization, operating systems, protocol compatibility, and NAS gateways.
By Jerome Wendt
Lower storage requirements, reduced legal liabilities, centralized management, and higher levels of data protection are all benefits that a consolidated and centralized NAS implementation delivers.
Yet as IT organizations consolidate their file-serving environments, new challenges emerge. Each NAS product might support different file systems and operating systems, which may impact the ability to integrate with technologies such as Microsoft’s Active Directory. Applications sensitive to cost and/or performance may not respond well when introduced to shared storage, change control, and central administration. And with the data of some business units and individual departments more sensitive than others, the ability to take advantage of features that centralized NAS introduces, such as global namespaces or shared storage, may not always be an option.
As users enter the world of consolidated NAS, there are some key questions that they need to answer as they transition from storage islands to centralized storage:
■ Can the consolidated environment re-use existing NAS appliances and/or SAN storage devices, or will it require new systems?
■ Will it need to support heterogeneous or homogeneous storage devices?
■ How does the NAS operating system support the Microsoft CIFS protocol?
■ Can the consolidated environment manage block and file traffic on the same device?
An initial factor in deciding which type of consolidated NAS infrastructure to implement is determining whether existing NAS appliances need to stay or go.
Re-using existing NAS appliances in a consolidated environment takes users down one of a number of paths. Users with multiple NAS appliances from the same or different vendors may find that a file virtualization appliance best fits their needs. A virtualization appliance identifies the files on each NAS appliance and builds a master catalog of all of the files. Once the indexing of all of the selected NAS appliances is complete, clients are re-pointed to the file virtualization appliance from the NAS heads that they previously used. The file virtualization appliance then redirects the file requests back to the NAS appliance where the file resides.
File virtualization appliances also facilitate data migrations from one NAS appliance to another without user interruption. This feature allows administrators to either move files to appliances that have more capacity or migrate data from existing NAS appliances to new ones.
Yet users need to take precautionary steps before they introduce file virtualization appliances. A key consideration for shops needing high availability is to make sure that file virtualization appliances offer configurations that provide the appropriate levels of uptime and availability, especially in environments that require high levels of availability and/or performance. For instance, NeoPath Networks’ File Director allows users to configure appliances to fail-over from an active appliance to a standby node non-disruptively. Other users may need appliances, such as Acopia Networks’ ARX6000 file virtualization switch, that give users the ability to perform load-balancing between nodes based on capacity, round-robin, or service-time options.
Another desirable feature is the ability to discover and report on the existing NAS environment before making any changes to it. For example, Attune Systems’ Maestro provides users with three levels of file virtualization: discovery, native, and extended mode. Attune’s vice president of marketing, Dan Liddle, recommends that users start with discovery mode when implementing Maestro.
“Discovery mode shows all of your storage resources and allows users to do some tracking and analysis. This gives users an understanding of where their file data resides, how it is accessed, and who or what is using it-without making any changes to the environment,” says Liddle.
Moving from the discovery phase to using the native file system offered by the file virtualization appliance introduces another set of risks. An initial consideration is how and when the files on the NAS appliances are indexed and created in the new catalog that resides on the file virtualization appliance.
Ali Zadeh, NeoPath’s chief operating officer, suggests users should only start accessing files through their file virtualization appliance after it has indexed all of the files on the NAS appliances that they plan to virtualize. “In this way, the data remains in a consistent state,” he says.
Another factor to consider is whether certain files may be included or excluded from the file migration. Once file virtualization is implemented, any servers accessing files virtualized by the virtualization appliance need to direct the requests to the appliance and not directly to the NAS head. If servers directly access files still on the NAS head that are already virtualized, data corruption could result. Scheduling this so file corruptions do not occur may require that administrators make the appropriate changes to server mount points for these directories and introduce new VLANs so servers cannot access both existing NAS heads and file virtualization appliances simultaneously. Either of these changes may necessitate taking applications offline temporarily until the change is implemented.
Users with all Network Appliance NAS appliances have another option to weigh. NetApp’s new Data ONTAP GX software allows users to scale out multiple NetApp filers so that they appear as one logical NAS appliance. When file requests are made to a NetApp filer that does not physically house the requested file, the filer redirects the request to the filer in the cluster holding the file. However, this feature is only available on Net-App’s FAS3050 and FAS6070 models with the new Data ONTAP G7 operating system. This precludes existing NetApp users from taking advantage of these features unless they either upgrade or replace their existing filers.
Salvaging SAN storage
IT organizations that need to salvage existing SAN-attached storage (either Fibre Channel or iSCSI) should consider NAS gateways (or “NAS-SAN gateways”). Like file virtualization approaches, NAS gateways reside between servers and storage, but do not attempt to re-use existing NAS appliances. Rather, all files are migrated from existing NAS appliances to the NAS gateways, which connect to storage arrays using block-based SAN protocols and present files to clients using the CIFS or NFS protocols. All files are then controlled, managed, and secured by the NAS gateway. However, migrating to a NAS gateway presents its own set of challenges.
First, users migrating to a NAS gateway may need to implement a file virtualization appliance to migrate files from existing NAS appliances to the new gateway. File virtualization appliances allow administrators to preserve the security permissions while migrating the files with minimal or no disruption to end users. This also allows administrators to take advantage of the discovery features and document their environment so they can more thoroughly plan the migration to the NAS gateways.
Users should also verify that the NAS gateway will support the storage arrays in their environment. For example, EMC’s Celerra NSX gateways only support EMC storage arrays, such as the DMX and Clariion lines. However, EMC’s senior director of NAS marketing, Mark Greenlaw, says EMC is thinking about extending Celerra support to include other vendors’ storage arrays. “We’re finding fewer and fewer homogeneous environments and it’s evident that users want an open approach so we’re evaluating this option,” says Greenlaw.
A variety of other vendors, including NetApp, ONStor, and PolyServe, support both heterogeneous and homogeneous storage environments.
Although ONStor’s Bobcat NAS gateway supports heterogeneous storage environments, the company found that some of its customers did not want to separate NAS gateway and storage purchases; instead, they wanted to buy everything from one vendor at the same time. To address this need, ONStor recently introduced the Pantera clustered NAS product, which bundles both a NAS gateway and disk arrays with either Serial ATA (SATA) or Serial Attached SCSI (SAS) drives.
PolyServe’s File Serving Utility addresses the need to re-use both existing storage and servers. Unlike NetApp and ONStor, which require users to deploy proprietary NAS gateway hardware and operating systems, PolyServe certifies a number of different server and storage hardware platforms and allows users to choose either Windows or Linux as the underlying operating system for the company’s NAS gateway.
NAS OS considerations
The operating system available with either file virtualization appliances or NAS gateways is generally one of three different varieties: Linux, Windows, or proprietary. Each of these introduces certain issues that users need to weigh in their decision-making process.
One potential concern for administrators is interoperability with the CIFS protocol used by Microsoft Windows to communicate with NAS virtualization appliances or gateways. Products such as Attune Systems’ Maestro and the Windows version of PolyServe’s File Serving Utility, for example, offer NAS operating systems that are based on Windows. This design grants them native access to the CIFS protocol since it is included with the Microsoft Windows operating system.
Vendors with proprietary operating systems, such as EMC and NetApp, need to take other steps to support Microsoft’s CIFS protocol in their NAS operating systems. For example, these vendors have joined the Microsoft Communication Protocol Program (MCPP), which gives vendors full access to all of Microsoft’s CIFS documentation and helps them to ensure their implementation of CIFS matches the current Microsoft Windows release.
Windows-based organizations considering Linux-based consolidated NAS options, such as those from PolyServe and Hewlett-Packard, may need to exercise caution due to some dependencies on open source code. HP’s director of NAS, Harry Baeverstad, concedes that the Linux-based version of HP’s EFS Clustered Gateway uses Samba, an open source version of CIFS, to integrate with Windows environments. This results in some features trailing the current release of the Windows operating system. However, Baeverstad is not overly concerned by this lag in development. “We find that by the time most users are ready to implement new Windows CIFS features, Samba has caught up,” he says.
Support for the iSCSI protocol is another feature that users may want to weigh when they are considering a NAS consolidation product. Currently available on all of NetApp’s products, as well as on EMC’s Celerra NSX with the DART 5.5 operating system, iSCSI allows administrators to use the same management console to manage file directories and LUNs and present them using the same Ethernet interface. It also allows administrators to manage advanced functions such as snapshots and replication in the same way for both file- and block-based storage management.
NetApp officials report that users most frequently deploy file- and block-based protocols in environments where they’re consolidating their Microsoft Exchange and SQL Server block-based applications with their file services.
One final consideration when you are selecting a consolidated NAS product is the ability of the software to segregate user and application data. Since NAS consolidations often bring together different departments with competing priorities, maintaining a degree of virtual separation between user data and application data-and prioritizing how it is managed-may be a requirement, even if keeping it physically separate is not.
ONStor, for example, addresses this issue by providing administrators with the option to implement virtual servers. Virtual servers allow organizations to assign certain administrators to manage certain sets of data while using the same back-end infrastructure. A key benefit is that this allows administrators to set policies governing backup, security, and performance characteristics on each virtual server, which helps to prevent rogue applications from consuming CPU and memory resources needed by high-priority applications. “Virtual servers help to maintain management granularity while achieving physical consolidation,” says Jon Toor, ONStor’s vice president of marketing.
Although NAS consolidation is often justified through impressive return-on-investment (ROI) numbers, IT organizations should not dismiss the technical and political challenges associated with implementing a consolidated environment. However, with the number of options available for file virtualization and NAS gateway appliances, this is a course that users can successfully navigate, assuming that an adequate amount of planning is done ahead of time.
Jerome Wendt is lead analyst and president of the Datacenter Infrastructure Group research and consulting firm. He can be contacted at firstname.lastname@example.org.
Is it time to consolidate?
With NAS consolidation gaining in popularity, here are some problem areas to look for in your environment that can help you determine if it is time for you to pursue a consolidated NAS strategy:
■ Too many file servers-An initial tip-off may be if you do not know how many file servers you have in your organization or where they are. This problem sometimes surfaces when you are building a disaster-recovery site and one adds up the cost of replicating the file servers at a secondary site.
■ Files not backed up-Scheduling backups on multiple file servers and trouble-shooting failed backups can become increasingly time-consuming. NAS consolidation allows users to store all of their files centrally, but also take advantage of advanced features, such as NDMP and snapshots, which facilitate instant backups and restores.
■ Difficult to achieve regulatory compliance-When you are audited or asked to produce all files relating to a particular topic or owned by a particular user or group, finding those files on individual file servers can be almost impossible. With NAS consolidation, administrators can generate reports to show who owns what files and where they are. Products from vendors such as EMC and Network Appliance also allow users to implement write-once, read-many (WORM) functionality and keep files from being overwritten or deleted after they are created so their data is preserved.
■ Managing out-of-space issues-If you are spending a lot of time trying to find or free up additional space on existing file servers, file virtualization and NAS gateway appliances can allow you to better use existing capacity or more freely add different (and lower-cost) tiers of storage into your storage pool.
For more information on the technologies discussed in this article, see the following articles that have appeared in InfoStor:
““Welcome to the file area network (FAN),” August 2006, p. 28
“Analyzing scalable NAS architectures,” August 2006, p. 38
“Users moving toward NAS virtualization,” February 2006, p. 39
Also, check out the archived Webcast, “NAS and NAS File Virtualization,” at www.infostor.com.