Volumes of unstructured data, such as email, spreadsheets and other files created by personal productivity software, are growing much faster than structured data. As a result, demand for network-attached storage (NAS) devices is also spiking.
NAS operates at file system level -- usually either NFS or CIFS -- and serves whole files to users over the shared medium of the local-area network (LAN), while storage-area networks (SANs) deliver data at the block level (via Fibre Channel or iSCSI protocols) and are well suited to transactional storage such as database access.
When it comes to leveraging existing NAS systems, storage managers are discovering that newer NAS features are making high-end systems look increasingly like SANs. Network-attached storage has retained its core functions of providing file access storage capacity, but available features have expanded to include SAN-like block-level access via iSCSI and SAN gateways, as well as high-performance file access functions, such as multiple heads for access to multiple disks simultaneously, global file systems and clustering.
These NAS systems have Fibre Channel as well as Ethernet ports and, increasingly, support 10 Gbps Ethernet (10 GbE). While the technical difference between NAS and SAN remains the same -- file-level versus block-level access -- from the user perspective it's all starting to blur, according to Tony Lock, programme director at research firm Freeform Dynamics.
According to Lock, the biggest problem with today's data storage technology market is that similar functionality is essentially offered by multiple vendors under multiple names. "Outside of a few storage gurus in data centres, people just don't have the knowledge of which solutions fit where," Lock said. "Simplification of storage networking is the biggest need."
This is a key driver for NAS boxes, which are increasingly being sold as unified storage with a single point of management, unlike many NAS gateways. IDC's market research finds EMC has the biggest share of the UK network-attached storage market at approximately 42%, followed by NetApp with a little more than 33%. IBM garners third place with around 10%, while Hewlett-Packard (HP) and Dell trail the pack. IDC's UK sales revenue numbers show a rise of 9.2% in NAS sales in the third quarter of 2009.
Helphire chooses BlueArc Titan devices for easy data storage snapshots
Saving a file system snapshot used to mean unmounting a file system to ensure file consistency. But data storage managers can now avoid this and quickly snapshot a live production system without affecting file access.
Francois Greyling, infrastructure architect at Helphire Group, found this to be a key factor when it was time to upgrade his firm's NAS systems. "We do snapshots every hour on NAS data and replicate them every hour. Snapshotting means if you keep enough of them, going far enough back, users can restore their own files by looking through previous versions and this saves time for IT staff," he said.
But snapshots add to the problem of growing data storage consumption, so Helphire decided to invest in BlueArc's Titan NAS devices. Among the reasons for their selection, Greyling said, was their ability to grow or shrink capacity on the fly, to consolidate disks for improved utilisation and performance, and to offer "a good disaster recovery [DR] solution that is easy to invoke and has low RPO/RTO [recovery point objective/recovery time objective]."
According to Greyling, "our principle is that we use clustered heads and replicate that architecture to disaster recovery systems, mixing and matching BlueArc with NetApp. We also ensure that the backup load is on the DR system rather than the production system, which allows us a bigger backup window." The BlueArc Titan devices are also used for VMware on NFS and to serve SQL databases.
The key technologies that drove Helphire's selection were faster snapshotting, thin provisioning, scalability, performance and flexibility. A big challenge was data management, which was resolved by imposing quotas and charging for storage so that data owners manage their own shares. "We can't know what's important to users and what needs to be kept, so they need to manage it," Greyling explained.
Helphire now has the capacity to deliver on the IT department's service-level agreements (SLAs), which include delivering business intelligence reports first thing in the morning. Overall, however, it was the low cost per terabyte and the high performance that clinched the deal.
NetApp replication, snapshots just the ticket for LBA
Increases in passenger throughput and expansion of its route network and terminal facilities were among the key drivers for Leeds Bradford International Airport's (LBA) upgrade of its servers and data storage.
Problems with the existing estate -- consisting of physical servers with local, direct-attached storage (DAS) -- included underutilised hardware, high support costs and energy usage, ineffective DR and decentralised backup processes. Without improvements, said IT manager Adrian Rollins, the problems LBA faced as it grew would only get worse,.
The airport decided that VMware-based virtual servers with shared storage were the solution. Criteria for storage included high availability and replication for affordable disaster recovery, ease of management and flexibility.
LBA selected NetApp's FAS2050 subsystems with approximately 24TB configured as RAID-DP (NetApp's double-parity RAID 6 technology), with data deduplication and replication functionality. Rejected alternative vendors included Dell/EqualLogic and HP, who didn't offer features such as deduplication on production data and zero-performance impact snapshotting.
Among the benefits of the new setup is a complete disaster recovery solution. The NetApp FAS2050's FlexClone replication and deduplication facilities allow LBA to replicate its entire IT estate, operating system, applications and data using virtual machines (VMs). This provides "excellent RTO," Rollins said. "In the event of a disaster, we just boot the VMs."
The DR plan can be tested and provides recovery point objectives of less than four hours with minimal data loss, and a within-minutes recovery time objective. It also provides a test and development facility for applications and the virtualised infrastructure.
The main data centre now provides CIFS-formatted storage for Windows shares and, via iSCSI, the VMware environment on two ESX servers. LBA uses NetApp Snapshot technology to maintain three days' worth of backups online and then archives them to tape.
The new storage systems also enable thin provisioning which, together with data deduplication, has reduced storage requirements by 77%.
Challenges to the implementation in LBA's 24x7 environment included downtime resulting from the upgrade, the small time window for converting physical servers to virtual machines and then deploying them, and the lack of time for testing different versions of database and OS versions with NetApp's software prior to migration.
"Other than that, the planning phase was successful and the implementation went ahead with no impact on business," Rollins said. "The only improvement I would make, if any, would be to install a second controller in the production FAS2050 and cluster the heads for improved resilience and zero downtime for storage maintenance. However, we have covered the business for DR by installing a second FAS2050 at DR and configuring SnapMirror with VMware's SRM."
There are other benefits, too. "Over five years, the VMware-on-NetApp solution will reduce our server/storage TCO by 21% or nearly £78,000," Rollins said. "In four years, energy savings will exceed £19,000; that's about a 44% reduction."
The combination of snapshotting, replication, thin provisioning, and a disaster recovery environment combined with wider benefits such as reduced data centre space requirements and a smaller carbon footprint, were among the key benefits that led to the NetApp deployment.
This was first published in January 2010