Multi-tier Data Management

December 10, 2008

You could be forgiven for thinking that everything is on a downward curve at the moment with stocks and shares, house prices and sterling all on the slide. The one thing bucking this trend is a slightly different currency, the currency of the digital age: data. 

The Problem

The volume of data held on corporate systems is growing exponentially, fuelled by the increasing use of rich content, remote working and e-mail. Today over 100 billion e-mails are sent daily and the capacity of data centres has grown by a factor of nine in just four years to cope with client demands for ever expanding storage. Even medium-sized organizations are now storing and backing up electronic data measured in terabytes, not gigabytes.

A Different Approach

This massive demand for storage provides one of the key challenges facing IT departments – devising an efficient, cost-effective data management solution when you’re faced with a moving target. You can invest x now, but in 12 months’ time, the inexorable mushrooming of data could well outstrip your storage and tape-based or online back-up facilities; so worst case scenario is back to the drawing board for a new plan, best case is allocating yet more budget to feed the beast. That’s not to demean back-ups, which are, of course, basic IT good practice, necessary for compliance and a fundamental part of any self-respecting business continuity strategy. But it does throw into relief the fact that too often we are treating the symptoms – riotous growth – rather than the cause: a basic lack of rigour when it comes to data management.

The simple truth is, we’re guilty of storing too much. We then back-up anything and everything, which is an honest belt and braces approach that comes at a price. Storage is cheaper now than ever but there is still a not insignificant back-up cost attached, especially when demands can grow by up to 50% annually.

Moreover, one doesn’t need everything on every data volume backed-up to primary or secondary storage: it’s estimated that 70% of files and e-mails on the average system will not have been accessed for six months or more. Having said that, there’s also a small amount of data that’s absolutely critical for normal operations – such data needs to be replicated or backed up and accessible on an almost real-time basis. And in between is about 25% of data that’s accessed regularly and is key to the normalcy of everyday working.

Given these different grades or layers of data, given their widely varying value to the business, wouldn’t it be sensible to treat them differently when it comes to storing and backing them up? Diagram A outlines the concept of Multi-Tier Data Management (MDM), where a hierarchy is established and each data type is given the most apposite treatment: replication, back-up or archiving.   

Diagram A

The overall impact of MDM is that companies back-up less, archive more and secure ultra-critical to legacy data. By utilising a hierarchical storage service, companies can identify, categorise, police and manage their data based on type, age and value to the business.

Diagram B (before and after MDM) illustrates very clearly what happens when such a quantum shift in the dynamics of data storage is allowed to happen: with the majority of data moved to cheap tier 3 and 4 storage, and the remaining 20-30% placed in appropriate tier 1 and 2 environments, you end up with dramatically reduced primary storage and back-up volumes AND significantly enhanced Recovery Point and Recovery Time Objectives.

Diagram B

The benefits of MDM read like an IT director’s wish list: reduction of overall spend on data and information management; greater control over data growth by classifying and archiving inactive data; reduction of primary data and thus reduction of back-up data and back-up window, shortening Disaster Recovery times; increased data availability by lessening the load on primary servers and storage; and more resources, money and energy freed up to spend on ‘added value’ projects.

Denton Wilde Sapte Case Study

The existence of ‘another way’ and the rich potential of MDM was spotted early on by Neil Pamment, IT director of Denton Wilde Sapte.  DWS have already moved a good way down the MDM route with an initial restructuring into two tiers: managed back-up (Tier 2) for rapid restores and secure offsite, online storage; and managed archive (Tier 3) for removal of aged data off the system yet still with easy end-user access for recovery and compliance purposes. Both these managed services have been out-tasked to InTechnology, who already provide DWS with a managed MPLS network thus making it that much easier to deploy additional services over the same pipe.

Neil’s original dilemma will be familiar to most. ‘As a law firm, our documents are like products for other businesses. They are valuable to us and we have to be able to back them up securely and efficiently. A specific issue was that we had a huge amount of data in e-mails and attachments that had to be accommodated.

We wanted a system of rapid backing up and securing of data that fitted with our operational timescales. We were using traditional methods of backing up to tape, but this took up too much time and there were too many problems. Tape as a medium works, but we also need to retrieve documents as well and tape is no good for getting data back quickly.’

As well as looking for a solution that would help DWS back up increasingly large volumes of valuable legal documents and administrative data, Neil was also focusing on the on-going security and availability of those same electronic collections. Data had to be available for rapid retrieval in case of disaster recovery and to assure business continuity at all times. ‘The need to get our data to an offsite, secure location was a second key issue for us,’ explains Neil Pamment.

The DWS team had a good opportunity to tackle the twin issues of efficient back-ups and effective data restores when the IT infrastructure in their City of London head office was being re-configured.

Managed back-up could provide the rapid restores that Neil wanted for day-to-day operations and for disaster recovery, but the data volumes involved were daunting. ‘An important feature of the managed back-up service is that data is compressed before back-up, so that our rapidly increasing data volumes were reduced. However, we still had vast volumes to back up. We realized that by introducing MDM and a value-based data hierarchy, we could instead archive off a substantial proportion of existing data and thereby reduce the size of back-up volumes. By subscribing to both managed back-up and managed archiving services, we could achieve efficient back-up and offsite storage for our data at reasonable cost.’

Conclusion

By changing the way they think about data and evaluating it according to its import to and impact on the business, DWS has taken back control of a situation that threatened to spiral out of control.

Wildly escalating growth in electronic documents, and in particular e-mail, has now been tamed. Primary data volumes have been reduced by 50% with a corresponding improvement in server and storage performance; it is a similar story with back-up windows and recovery times, with durations decreasing by 50%. Mailbox management effort has been hugely minimized by archiving off inactive mail, which has in turn enhanced the day-to-day running of Exchange.

In short, DWS are saving thousands of pounds every month while taking advantage of a fully optimized data management solution: all end-user data is now policed and actively managed; all archive data is taken off-site and replicated between two data centres for extra protection; and all remaining primary data is backed-up daily, available for immediate on-site or online recovery.

From a two-tier data management solution it is but a small step to add a third, topping off the hierarchy with a layer of replication for ultra-critical data. While that may be seen as an indulgence by some, for those firms heavily involved in transactional processing it is a godsend: instant roll-back to the data of an hour ago may just be the difference between remaining a panel member or being summarily dropped.

In either form MDM has been designed to provide a very workable, cost-effective and efficient framework for greater data control, improved business continuity, heightened corporate governance, reduced vulnerability and exposure, and enhanced system and application performance. Like the downward trends in the economy, the relentless progress upwards of data volumes is bad news – and a big challenge. Here’s one way to cut it down to size.

Richard Quine is Director of Product Management at InTechnology: richard.quine@intechnology.com.