Want to see the future of storage? Look to Linux

Fashion can deceive, but backing Linux is looking a cosy fit.

Fashion can deceive, but backing Linux is looking a cosy fit.

Fashion often seems to rule the IT world, sometimes beneficially. It used to be said that you couldn't get fired for buying IBM. Then the trend became that you couldn't be fired for buying Unix, or Windows servers. Fashions change, and the safety net is now moving under Linux. This, however, could turn out to be a very good thing indeed.

Linux has several attributes that make it of particular interest in the storage world:


  • It's cheap - often free - and Linux skills are plentiful

  • It's free of the licensing restrictions that place access to other operating systems under the whim of an intellectual property owner n It's open to public scrutiny.

In the 1950s, science fiction writers thought the world would be run by a single massive computer. But reality has proved the opposite: the internet has no centre. Processing power in orders of magnitude greater than that of PCs is shipped in cars and cellphones. Even bus stops now have computers in them. Peripherals often have more processing power than the systems they serve.

This trend is set to continue, with more intelligence appearing at lower levels each year. Each unit will require an operating system, and this is where Linux might take on a future starring role.

With the increasing processing demands placed on storage devices by data virtualisation, it seems logical for processing power to follow cache into the hard drive itself. Basic functions such as error-checking code calculation are there already. Future storage subsystems may use thousands of processors. This requires an embedded operating system, for which Linux is ideally suited. Versions using as little as 150Kbytes have been devised, such as that for the Terapin "mine". PDAs and cellphones based on Linux have also been produced.

Having no intellectual property owner to say "yea" or "nay" on a whim frees product developers from strategic worries. Corporate fallings-out can destroy products and sometimes companies - as in the case of Sendo. Linux releases companies from such considerations and makes many investments safer, lowering the cost and risk of entry to the market and thus increasing choice.

Linux offers another great strength compared with other operating systems: it is robust, secure, and improving in both respects. IBM's z/OS is regarded as the most secure operating system on the market, in part because it is based on earlier code that was as open to inspection and criticism as Linux is today. Only when major changes were made to MVS in a closed source fashion - Unix System Services - did chinks appear in the armour.

Some governments, such as the Russian and the UK, have become so concerned that they have demanded access to the source code for other operating systems. But this monitoring cannot match the close examination to which Linux is subjected by the hundreds of thousands of programmers who have its source.

Dedicated storage subsystems with proprietary operating systems come under the closed source umbrella. The systems being announced by the major suppliers this spring may prove to be, at best, the penultimate examples of their kind. Perhaps we will see one more generation before they are all replaced by alternatives based on open source and the standards that are currently being established.

Management of storage capacity is both a problem well on the way to solution and much less of a concern: storage capacity is now outstripping data volume growth. Data management - especially in complex environments - is much more of an issue than capacity administration.

In the days of mainframe hegemony, database administrators dreamed of a "data dictionary" to describe an organisation's knowledge base, though few were ever implemented. But back then everything was proprietary and development platforms were few and expensive.

Today standards are open, accessible and can be downloaded from the internet, and development platforms are in every back bedroom. Data virtualisation techniques such as IBM's Storage Tank can supply a form of data dictionary. Its metadata servers are already Linux-based.

The Linux environment is famous for sudden problem-solving. It doesn't seem to matter how complex the problem is, nor do we ever hear how many attempts at a solution failed.

It seems inevitable that Linux because of its low cost, scalability - especially downwards - and security will become dominant in future generations of storage products. It will bring all the advantages of the Linux world with it, including some surprises. So sorely needed data virtualisation functionality may arrive sooner than was previously thought, and it will certainly be more affordable.

Phil Payne is managing director of Isham Research and one of the UK's most respected analysts for storage and enterprise systems


Read more on Computer storage hardware