Construct a smarter home for your data

Manchester-based Ainscough Crane Hire has lifted storage performance by backing up data to a disk-based system. Pic credit: Flickr: IRP IMAGES

The chances of any IT department reducing the volume of data created by its users are almost nil. Growing file sizes, an increased reliance on email, and regulatory pressure to retain information for a set period, are just some of the factors that have conspired to create a seemingly insatiable appetite for additional storage capacity.

But there are ways to optimise existing storage resources to reduce the amount of idle hard disk space, and to lower costs by introducing more energy-efficient technology that uses less power and helps with in-house disaster recovery and business continuity plans.

One such technology is solid-state disk (SSD) storage, which provides faster access speeds than even the fastest hard disk drives (HDDs). Rather than spinning disks, SSD uses a Flash-style memory chip that holds data even when the drive is powered off.

SSD has far superior data access capabilities that provide greater efficiency and performance for frequently-accessed, mission-critical applications such as search engines, financial and online transaction processing, web servers, databases and video on demand.

Although the initial purchase price of SSD is far higher than equivalent capacity hard disks, there is significant return on investment potential, says Garner research director Joe Unsworth.

“This potentially requires fewer servers, fewer software licences, and often results in return on investment that pays back the higher initial expense in a few years depending on the application and workloads involved,” says Unsworth.

SSD also consumes considerably less power than HDD technology, but it is important to remember that not all vendors’ SSDs are created equal.

“Power efficiency is beneficial since SSD contains no moving parts and does not require cooling as would be the case for fast HDD technology that can get hot, but this is difficult to quantify because it depends on usage, workloads and other factors,” says Unsworth.

Arguably, a surer way to make data storage more energy efficient is server and storage virtualisation. Many organisations have found that replacing large numbers of physical servers, each drawing their own power supply, with multiple virtual servers that run as images on a single piece of hardware, saves money and space that can be put to more productive effect. And connecting those virtual servers to hard disk arrays through a storage area network (SAN) also helps.

Ian Booth is IT manager at Ainscough Crane Hire, which operates a national UK network of 30 depots employing about 1,000 people. Until recently Ainscough was running 15 physical servers at its head office in Manchester, with around 350 IT users accessing SQL Server databases and file and print services.

Faced with rising costs and a growing volume of data distributed across multiple physical servers, the company decided it needed to implement more efficient storage and backup systems that could also help the firm to lower its IT costs.

“We were looking for a small SAN and the project cascaded to the point where we did a lot more with data security, backup, resilience and extended our disaster recovery platform to a secondary site in Standish,” says Booth.

The firm installed VMware’s ESX server virtualisation software on two physical servers, and an EMC Celerra NS20 unified storage system to host its data. Using SnapView to create instant snapshots of system status has also removed Ainscough’s reliance on backup tapes from which data was harder to find and restore.

While Ainscough is one of a growing number of organisations that have decided to move their backup from tape to disk, tape-based backup is still widely used in many other organisations.

According to Hamish MacArthur, analyst with MacArthur Stroud International, tape still represents the most cost-effective data storage medium for many companies, depending on their service level requirements and the type of data they want to store.

“Tape will be around for a long time to come and is energy efficient as well,” he says. “It is suitable for data archiving if you have to keep something for 20 or 30 years but do not require it to be regularly accessed.”

Virtual tape libraries – ­ tape backup drives that appear to the SAN as attached hard-disk resources ­ – are widely used. But they are just one element of a broader storage virtualisation strategy that involves capacity from multiple physical storage resources, either direct attached (DAS) or network attached (NAS), being pooled together as a single, centralised entity transparent to the application accessing it.

The key question is how to go about provisioning and managing those virtual resources, says MacArthur.

“It is about how basic capacity provisioning for live applications and utilisation of users’ data storage capacity can be done more efficiently, but there is still some way to go on the information management side,” he says.

Hierarchical storage management (HSM) and information life cycle management (ILM) software go part of the way. These are applications that move data between different tiers of storage resources ­ – ranging from high-performance, high-cost Fibre Channel disk to lower cost, slower SATA hard drives or tape, for example –­ automatically and according to compliance and data protection regulations that define how long data should be stored for and how quickly it should be accessed.

For provisioning and capacity utilisation, two technologies that can play a significant role are thin provisioning and data de-duplication. Thin provisioning limits the amount of storage capacity lying idle waiting for an application or user to fill it with data by allowing the server to view more storage capacity than is physically available –­ up to a certain point, fewer physical hard disks are required to handle anticipated data expansion on a shor t-term basis.

Data de-duplication has a simpler remit, which is to eliminate multiple copies of the same data stored across the SAN, keeping a single master record that is accessed on demand. With de-duplication there are huge savings even with a 20:1 compression ratio that delivers a five per cent storage utilisation, and many are getting even greater compression than that,” says MacArthur. “The challenge is to apply it in a consistent way across the infrastructure.”

Return on investment from storage upgrades can still be an elusive goal though, not least because having so many elements makes it hard to quantify. Ainscough originally estimated that storing data on virtual servers, for example, could save £10,000 per year in lower power costs, though Booth says he has not noticed any significant reduction yet.

“The really obvious saving is on the disaster recovery contract with all our business-critical software now on a backup server,” says Booth. “Otherwise, we have not yet identified any cost savings. We spent a couple of weeks going through our power bills, but we still use the same air conditioning so there is no change there. It is more about reliability and peace of mind.”

A similar story comes from George Nesbitt, general manager of business development and IT at Newcastle International Airport. He recently installed two small SANs connecting two virtualised servers at two separate sites within the airport to aid disaster recovery and backup.

Each SAN has 2TB of pooled data capacity, though Nesbitt says only 1TB is currently in use. The systems support up to 80 regular users at any one time, with the airport employing around 250 shift workers to keep its systems running 24 hours a day.

The two servers share an application processing load that supports a Microsoft Exchange email server, file and print services, the airport’s payroll system and its web server. As mission-critical systems, these are replicated across an IP link that runs over a fibre-optic network recently installed on site.

“We used to use an external company to provide our disaster recovery but that has now been brought in-house so there is a saving there,” says Nesbitt, who spent £200,000 on a DataCore SanMelody system that forms the basis of the SAN, including the first year’s support and maintenance charges.

Finally, it can cost less to use storage-as-a-service or storage-on-demand offerings now available from ISPs and cloud computing providers. Much like similar small-scale services for consumers and small businesses from the likes of Google and Amazon, these offer hosted storage capacity backed by value-added services such as backup, disaster recovery and replication.

In part four, we get the experts’ views on the legal and technical aspects of storage

Best Practice | Andrew Reichman

Five tips for storage efficiency

For some time, organisations have spent more time building reliability and performance into their storage environments than they have on cost efficiency. But with budgets shrinking and growth continuing unabated, the squeeze is on to cut costs in storage.

Prioritise measurement and reporting. You cannot manage what you cannot measure, but it is amazing how little maturity there is in storage reporting. Firms have been willing to spend millions on capacity, but have been fairly unwilling to spend on reporting tools. Storage environments need accurate and fairly automated reporting tools. These can be custom-developed tools or off-the-shelf storage resource management products, but either way, a tool is needed to have effective visibility to support the change required to improve efficiency.

Use more dense drives. It is not the sexiest cutting-edge technology, but SATA drives hold 10 times more capacity than Fibre Channel, and they cost about the same. Any way you can put more data onto dense drives will dramatically improve the cost structure of your storage environment. The two main ways to increase use of dense drives are tiering and system improvements such as wide striping that allow SATA drives to be used in production for performance-sensitive data.

Reduce the storage footprint. Experience suggests the industry average for storage utilisation is between 20 to 40 per cent. Traditional storage systems require a fixed allocation up front, and changing it is cumbersome, so users often ask for far more than they need to prevent delays in the future - this leads to significant wasted capacity. Storage reclamation can help to put unused capacity back into the free pool, and thin provisioning can help prevent it from being wasted in the first place.

Right-size performance. Smart firms looking to shed storage cost have started to become more granular, delivering just enough performance to meet business requirements, but no more. A services catalogue that defines guaranteed performance levels and a chargeback mechanism to assign a price tag to services help to raise cost awareness and give a tangible motivation for business to use cheaper stuff.

Focus on simplicity and consistency. Many firms like to select best-of-breed products for each aspect of their environment, but end up with a multi-vendor environment that is complex to manage. Take the bigger picture and plan for a consistent environment that is easy to manage. Select a small number of vendors and design standard configurations using their products that reduce the overall complexity.

Andrew Reichman is a senior analyst at Forrester Research.

Please visit www.forrester.com/computinguk for several complimentary reports made available to Computing readers by Forrester Research.