It’s simply impossible to avoid the V-word in any discussion of data storage these days: ‘virtualisation’ is a six-syllable mantra mouthful that pervades all vendor literature and techy debate. Naturally, this being the IT industry, a lot of that debate is about what it actually means! Another theme is whether or not it has actually been delivered anywhere — especially in an Irish context. Virtualisation is a concept or approach to the management of data in a computer system that has in fact been around for a long time, originally in the mainframe and mid-range environments. Since it pre-dates our more recent Internet age ‘virtuals’ like private networks and reality — and has only begun to be applied to the mass market for high volume data storage since the latter end of the Nineties — it’s probably not surprising that definition is a bit mushy.
All of the major storage equipment manufacturers and many smaller specialist outfits have now jostled their way onto the virtualisation bandwagon. The highly respected Storage Network Industry Association, which should not really be described as a bandwagon but is not the conductor/bandmaster either, has over 100 member companies. Even SNIA itself acknowledges: ‘storage virtualisation has become the buzzword in the industry, especially with the increased acceptance of Storage Networks. But besides all the hype, there is a lot of confusion, too. Companies are using the term virtualisation and its characteristics in various and different forms’.
You betcha! Virtualisation is a buzzword in the real IT tradition — simple in concept and thoroughly confusing in its many manifestations. Basically, the idea is that you manage your data storage resources as if everything were one common pool and to do that you make all of the separate storage devices and bits and bobs effectively invisible from that control level. Only the underlying storage management system needs to be aware of what’s on Location X, Device 1, disk 3, etc. Taken a step further, you group the data by human or business logic — Donegal Customers, Key Accounts, In-company Email, VAT or whatever. The importance and the purpose is that users and systems administrators do not need to be IT specialists and can wield everything easily. The organisation makes the best use of all of its available capacity — essentially because everything is automated and smart software makes the decisions based on policies and rules that have been set up initially to suit the specific requirements of that organisation’s network and its traffic. It only needs to add storage when the total pool is running low, as opposed to any single area or device. It can also, the theory goes, use inexpensive commodity disk drives.
Gritty reality
As might be anticipated, virtualisation is turning out to be another great vision encountering gritty reality. For a start, the V- approach can be applied at the appliance level (disk array), at the host or server level and at the network level — which can also mean enterprise level. Add to all of that the fact that different vendors offer their individual virtualisation solutions and you have a really confused market. In fact some major vendors now offer different virtualisation solutions at different levels. Sceptics might even suspect that since virtualisation is the in thing, but not too well defined, it is easier to stretch the definition than actually develop or design virtualised systems from scratch.
Then there is the fact that virtualisation should properly be associated with open systems, logically and because that is what the storage sector of the industry is promoting as a major advantage: if your data can be viewed and managed as a single transparent common pool, separate from the storage hardware, then the breed of the devices involved should not matter. This is particularly important in today’s cost conscious market, where discarding legacy devices is not an option.
Virtualisation is seen as the coming technology in data storage management, but as Tom Keane of CMS Peripherals points out: ‘the important thing is the management. With the ever-accelerating growth of data, organisations have to look closely at the ways in which they handle it — and the hidden costs of doing so badly’. As enterprise sales manager of CMS, which handles storage and related products from over 50 manufacturers, Tom Keane says there has been a real surge in data volumes in recent years. ‘About 40 per cent of Irish business spending on IT hardware is now storage-related. Many industries have compliance issues that require data to be retained for longer and email is actually a data storage problem for many businesses. But in fact we are just generating more electronic data. It is what you can do with that data — and do easily — that actually counts. Some needs to be highly and speedily available, some can be archived, mission-critical stuff may be replicated and all of it has to be backed up.’
These are the factors that are driving the adoption of Storage Area Networks (SANs) in Ireland and, although it is something of a cart/horse situation, that growth of the SANs market is what is dragging virtualisation in its trail as an effective approach to the Storage Resource Management (SRM — yes, God help us, another ambiguous acronym) issues that arise. ‘There are specific technical issues with virtualisation and differing solutions,’ says Michael Coyle, IBM’s storage products manager. ‘But from the users’ point of view you could see it as one huge, unified SAN-wide file system. You can see your data clearly by normal filenames and directories, organised logically to suit the way the organisation is run and its applications. But behind the scenes the data is in fact organised to take advantage of the technology. In the meantime, every single file has attributes that are set according to user-defined policy and determine how it is handled, such as the appropriate level of availability or the degree of security.’
‘Dashboard’ concept
IBM is one of the world leaders in data storage, rivalled in this as in other markets by Hewlett Packard and the Dell-EMC combination which share the mid-range market between them. In Ireland, that is the vast majority of business with serious data storage requirements. Paul Marnane, HP enterprise marketing manager says the company has numerous virtualisation implementations in Ireland, with the Department of Finance at the top end and many examples of virtualisation at the disk array level. ‘It’s all about the move towards automation and storage management software to make cost-effective deployment easier. There’s the “dashboard” concept where users can see easily what’s going on and organise their resources for the needs and priorities of different departments. It might be live, instant data for a busy call centre or all the way across the spectrum to off-line storage of seldom-accessed material. It is always dynamic — if priorities change, you just change the rules and if more capacity is needed you just add it to the pool and the system self-tunes to take it on board and manage it.’
EMC is the largest storage specialist vendor in the world and thriving on the explosion of data generation in recent years from the Internet, rich content (images and multimedia) and large enterprise applications. ‘More and more capacity required — great!’ is the happy response of UK and Ireland marketing director Nigel Ghent. ‘Email is another major element, which has really become mission-critical and has major associated issues like compliance and security. Compliance means that keeping all sorts of records for long periods is now mandatory in most countries. In simple terms, the only problems with all of this are management ones. We want to make good use of our data — and get hold of it very quickly at some times. But particular bits of data have a life cycle and drift towards the archive end of things.’ What is basic to storage management, Nigel Ghent says, is the appropriateness of the platform to the importance of the data and its stage in the life cycle. There is no inherent problem in having absolutely everything instantly available — it will just cost a lot of money. ‘But it does seem to be getting to the point where most information is required fairly quickly — or at least that is the expectation — so we have the idea of “near line” data storage and a blurring of the distinctions between disk, tape and optical devices. New technologies will come along — holographic, solid state memory, for example. But EMC is actually agnostic as far as the storage medium is concerned. We’ll always go with the mainstream and the right cost/performance balance. The important thing always is that the only reason we hold data at all is to extract value from it. Today’s storage management is all grown up and the emphasis has switched from the devices to the control software — it’s all about leveraging that value to the maximum for whatever the organisation or business does.’
The sheer importance of the formerly dull and boring data storage niche market today is exemplified by the emergence of specialist companies like Data Durability, founded just last year as an independent storage solutions company. ‘The trend towards SANs is inexorable,’ says Alan Yaverbaum, storage consultant, ‘and very simply the key to successful management of a SAN is virtualisation. It simplifies things but it also means that administrators handle the attributes of storage, which is what relates to the needs of the organisation, rather than the mechanics. The right solution can offer a greater return on investment through savings on labour and greater hardware utilisation and uptime. It is also important for a truly scalable SAN, which involves more than just capacity increase, but hardware and topology growth and change’.
Another company that has developed a major specialisation in data storage is Unitech, IBM’s largest reseller of mid-range systems. ‘Storage resource management has just loomed larger and larger in our business,’ says director Tomás O’Leary, ‘because the clients are increasingly concerned with it. Even the daily routine of backing up to tape from each server and rotating the tapes requires too much labour for many businesses — and with all that human intervention there’s bound to be some sort of cock-up occasionally! So in fact the easy facility for centralised backup is actually a serious selling point when organisations consider a SAN solution. It comes along with the other benefits of consolidation, costs savings and better data management — whether virtualisation comes into the picture or not’.
Value and risk
He emphasises that the first and most important thing in managing data is to differentiate it according to the value to the organisation and the risk posed by loss. ‘All storage management is built around protection and availability, with a lot more electronic data being mission critical because there is no paper equivalent. We prioritise according to how critical the information is. Once a company begins the process of storage consolidation for all of the obvious reasons it is not a huge extra step to go to smart differentiation. In a SAN it is then easier to manage the storage resources according to policies that follow those priorities. The clincher is that the total package is very cost-efficient and the RoI picture is excellent.’
Whatever about the specific data storage architecture, it is interesting that all of the experts we spoke to pointed to two factors as the most significant reason for the rapid growth in data storage needs: new types of data and an expanding range of compliance requirements. Email is the most often cited single type of data that has absolutely exploded in recent years. But there are also ‘rich content’ files, starting with simple colour in previously mono documents then going through desktop publications and html Web pages, colour images and multimedia. Lester Horsfall of Sharptext pointed to a whole new genre — security CCTV video transferred to disk. It is a secure and cheap solution for pubs, clubs and entertainment centres generally, greatly encouraged by the insurance industry which likes such records kept for at least six months.
With more orthodox business data, the major driver is the fact that our No. 1 technique for security against loss is to make copies of everything, whether replication of production system data for instant failover, or off-line or off-site business recovery, or simple backup. Amazon.com, for example, is reputed to make multiple copies of its entire database every 15 minutes. But regulatory compliance is a major factor, pervading many types of organisation and contributing hugely to the rapidly growing data mountains. Governments are requiring telcos and other public services in the interests of security to retain call records etc. for minimum periods of several years. Auditors and financial regulators are now red-hot on electronic records, stoutly guarding slammed stable doors long after Barings, Enron, Allfirst and the others have galloped into the sunset. In the pharmaceutical and chemical industries, the US Food & Drugs Administration all but notorious ’21 CFR Part 11′ requires manufacturers to record and retain batch data (from perhaps hundreds of sensors) until two years after the expiry date of the products! That alone is driving multi-terabyte data storage in all of the companies in the industry — and has created major business opportunities for Irish software company, Automsoft, which automates the process.
So we can now add a third life certainty after death and taxes: mankind will always create more data. The data storage resource management niche is rapidly becoming a major sector of the IT industry, now gaining the kind of technological impetus hitherto reserved for go-faster chips and handheld devices. It does seem clear that ever smarter, ever cheaper commodity disk storage will see out the decade as the principal medium, with some analysts predicting the demise of tape media for all but pure archiving. Right now, the SAN approach to the architecture and virtualisation in some of its many colours as the management front end are the clear race favourites. But it would be a brave hack or analyst who would predict anything with certainty or even suggest whether we are on the back straight or just coming around a bend.
Data Storage Jargon
A Storage Area Network (SAN) is a dedicated network of storage devices, separate from the LAN or WAN. It uses optical fibre for high speeds and makes all data storage available to all servers anywhere on the network — they really only act as pathways for data from SAN to end user. The ‘network’ element of a SAN can be misleading, since the thing might well be a single box — or a multi-location network of devices. Since data does not reside directly on any network server, the power of servers is utilised for business applications and network capacity released for those applications. Storage capacity and additional devices can be added without disturbing the network. Although SAN is essentially a hardware architecture, specialised software to manage all of the data traffic and different levels of priority, availability, etc. is an integral part of the solution in the organisation.
Other terms in regular use are Direct Attached Storage (DAS), essentially the traditional hard drive(s) in the network server or even specific PCs together with tape backup devices. Network Attached Storage (NAS) is a newer term for what is, essentially, a server that is dedicated to file sharing. It can exist anywhere in a LAN or be made up of multiple linked NAS devices. The traditional network server still handles all of the processing of data and functions such as e-mail, authentication or file management, but the NAS device delivers the data to the user. A key feature is that NAS allows hard disks to be added without shutting the network down for maintenance/upgrades.
23/09/2003
Subscribers 0
Fans 0
Followers 0
Followers