Whats The Fuss Storage Networking
What’s the fuss… about Storage Networking?
A few years ago, some bright spark noticed that storage devices – disks, tapes and so on – were tied too closely to the computers in which they were installed. What if, he (or maybe she) thought, what if storage devices were given a network of their own, or even attached to the LAN directly? All storage devices could be shared between all the computers simultaneously, new storage could be added or replaced at any time, and information could be managed, moved, backed up and secured from one central place. Nice dream – this is the vision for storage networking, in all its forms. Networked Storage, or the networking of storage, involves creating an environment in which storage hardware (in its simplest form, hard disks, tape drives and so on) can be directly attached to a network of some form. There are two ways in which this can be done:
Create a network (of protocols, devices etc.) exclusively designed for, and hence optimised for, storage. Storage devices of all forms can be arranged in a form that best suits storage needs, for example data can be mirrored to a fail-over site. Also many data transfers (say, backups) can take place without taking bandwidth away from the regular network, or processor time cycles from servers. This model gives us Storage Area Networking, or the SAN.
Create devices which uses regular network protocols (usually Ethernet) but which are built from the ground up to store and retrieve data. Such special-purpose “appliances” can be optimised for performance criteria such as data throughput, reliability and information integrity. At the same time, they can be built more cost-effectively than general-purpose devices (regular computers to you and me). This model gives Network-Attached Storage, or NAS.
The similarities and differences, costs and benefits, strengths and weaknesses of SAN and NAS have been debated beyond the call of duty within and outside the storage industry. Let’s face it however, most people couldn’t give an orangutan’s elbow for the differences between SAN and NAS. In the future, it is fully expected that the two models will merge; meanwhile, we are blessed with both.
The business benefits of storage networking may be summarised in one word – value. Correctly specified and implemented, a pure storage environment is able to deliver a better, faster storage service at a lower cost. With NAS, there is less of an impact on the existing network infrastructure, physically at least (however it is recommended to take account of the additional network load of adding new devices). SAN involves a greater up-front investment – there is a new network to be implemented, after all – but boasts greater advantages of security, performance and resilience. As with any other infrastructure deployment, storage implementations are as much about about understanding (and therefore meeting) the business information management needs of the organization, as having a grasp of available technologies and how they should be deployed and maintained. Don’t believe anyone who says, “you just plug it in and let it run.” Keep in mind the following:
The most fundamental storage management application is (still) backup and restore, with the most important being restore. Addition of storage should equate to the addition of managed, fault and disaster-tolerant facilities, not just plugging in of disks.
Interoperability remains the bane of the SAN, as it has been for the past three years or more. Relatively new standards, such as those from the SNIA, are being implemented to ensure both hardware and software interoperability, but it is still a brave IT manager who invests in SAN storage hardware from different manufacturers.
Scalability of storage should be a problem solved, but it has to be implemented. Technologies such as storage virtualisation are being discussed and implemented as ways of managing and using storage assets; however the definition of virtualisation varies from manufacturer to manufacturer, from virtual physical ports on hardware, through virtual placement in a silo to all storage assets being viewed as a single, infinitely expansible pool.
Where’s storage networking going? Virtualisation is a no-brainer, at least it will be once the manufacturers have agreed what it is. Certainly, from the customer perspective it would solve a lot of ills, such as the old chestnut of how to manage the disparate, distributed, duplicative, fragmented storage environments that exist in many organizations. Virtualisation cannot exist without suitable management tools, which may exist for on a manufacturer by manufacturer basis, but are still in their infancy (consider CA’s BrightStor portal and HP Openview Storage Node Manager) when it comes to really getting a handle on an organisation’s storage assets. The next evolutionary step, following “virtual” and managed” is the provision of storage as a managed service: this has many dependencies ranging from the physical – available bandwidth and implementation of caching technologies – to the emotional – would you really trust a third party to manage your data?
Whatever storage insiders would tell you, there is still a little way to go. Technologies are improving all the time, new standards such as iSCSI are being pioneered and each new step is further approaching but yet delaying the vision of coherently managed, fully multivendor, networked storage delivered as a service. Mustn’t grumble – technology is only part of the problem and the savvy IT manager would do well to use the time available to understand his storage needs and how well they are being met. So – is your storage house in order?