Storage: the inside story

By
16 September 2002 12:00 PM
Tags: internet, ip, magazine, networks, servers, convergence, attached, technology and business


The most obvious ongoing trend in storage is simple: size. Storage capacity, whether it be in disk, tape, optical or other formats, is steadily increasing, and the cost per megabyte continues to fall.

Don’t be fooled, however, into thinking this means that storage is getting cheaper. “The vast majority of the cost in storage is not in the disk but in the personnel,” says Simon Johnson, director of enterprise solutions at Dell. The problem is exacerbated by businesses which choose the seemingly “cheap and quick now” option of simply adding extra servers when storage demands increase.

“They have farms and farms of servers to manage, and it’s not a very elegant solution,” says Harry Christian, regional director for Network Appliance. “They’ve grown on the back of storage needs, rather than using servers to deliver applications.”

The trend is understandable. “Many applications need lots of temporary space, so people tend to overprovision, and you end up with excess,” says Graham Titterington, senior analyst at research company Ovum. But such an arrangement is a nightmare when it comes to planning backup scenarios. “Any kind of disaster recovery based around multiple servers or multiple islands is very difficult,” explains Johnson.

“There are lots of organisations with a variety of storage solutions. How do they know their whole environment is being backed up?” asks Richard Collins, head of Computer Associates storage division.

Everyone agrees that widely dispersed servers and storage are a horror story waiting to happen. “If you have a lot of point solutions out there, it’s all too easy to have different policies on them. There needs to be some consistency,” says Grant Smith, storage manager at Tivoli. Greg Bowden, national business manager for integrator Dimension Data, agrees: “You can’t control a large infrastructure with point solutions all over the place.”

Nonetheless, for a variety of reasons, capacity demands continue to rise. “People are frightened to delete data,” according to Joan Tunstall, marketing manager for StorageTek. “It may have some business value that you don’t perceive today. If it’s something like e-mail, you may get into trouble for deleting it.” (Just ask the guys at Enron or Microsoft.)

So how can you save on personnel costs and work more efficiently? Make things simpler by consolidating. “Consolidation is something that has really become top-of-mind for many businesses,” says Christian. “It’s not only about server consolidation, but bringing everything together. It’s not about the upfront costs only.”

Others agree. “The umbrella trend is simplification and consolidation, bringing together islands of storage from servers and desktops,” says Johnson. “It’s a pattern that’s going to continue for some time.”

“If you consolidate into a central pool, you only need one lot of working space,” offers Titterington. “Of course you can just buy more disks, but then you have a big management problem on your hands. You get to the point where it takes longer to do the backup than there are hours in the day.”

Hard-headed planning is critical. Businesses are asking questions prior to authorising purchases of hardware or changes in infrastructure. “Lots of CIOs are requesting more information on the types of data that reside in the organisation,” says Collins. “We’ve also seen huge growth for running storage resource management services.”

There are limits to the process of consolidation, but even if dispersed hardware is unavoidable, it can often be better managed. “It may not be physically possible to consolidate all your storage, but you can at least unify your storage management,” says Smith.

Network it baby, yeah!

“Storage technology is in some ways a very immature market, because there’s still a high rate of innovation,” says Clive Gold, ANZ marketing director for EMC.

Yet that innovation is not always in areas of high customer demand. “Nobody’s knocking down our doors and saying they need a faster subsystem or more capacity,” said Gold.

The main area of development in storage hardware in recent years has been the move towards networked storage solutions. “Organisations are looking for a more holistic view of their storage assets, and are looking to connect islands of storage with new, lower-cost interconnect options,” says Simon Elisha, senior systems engineer for Veritas.

Overriding all of this is the need to make the management of all of this storage cheaper. The move to networking is seeing a decline in simple direct attached storage, though not to the point where it is becoming obsolete.

“I don’t see direct attached storage dying out, but people will be networking storage devices almost universally,” says Titterington. In Johnson’s view, “Direct attached storage still has a place, in test environments and in standalone storage.”

Another important market is in notebook users, who have unique backup needs.

Once committed to a networked system, the dominant debate in recent years has been over two approaches with disturbingly similar acronyms: storage area networks (SAN) and network attached storage (NAS). Storage area networks allow any-to-any connections between servers and back-end storage, blurring the distinction between individual devices, while NAS simply makes storage devices network accessible.

While vendors have promoted both technologies as rivals, the truth is a little less dramatic. “NAS versus SAN—that argument is going away,” said Christian.

In Elisha’s view, “SAN is by far the most effective solution for most enterprise storage requirements, particularly for database applications. NAS has a place as a file server on steroids, but the rush to host all applications on NAS devices is a little premature.” “SAN and NAS are potentially quite complimentary,” says Titterington. “SAN is a very bulk, heavy-duty, low-level type service. Once you’ve got it up and working, it’s robust, reliable, and can handle large volumes of data. It’s mostly used within a data centre or within one site. You’re looking at LAN scale rather than WAN scale. NAS works at the file level and can connect to any other network. You can just use Ethernet for it if you’re not doing particularly high volumes. And you could have a SAN behind it; NAS can be used as a gateway into the SAN.”

SANs are also working their way further down the enterprise food chain. “I don’t think company size plays a part in choosing a SAN—resources, infrastructure, and budget are the key indicators,” says Johnson. That doesn’t mean that either family of products are ready for true mass-market adoption. “The current products are not very friendly,” according to Titterington. “There’s a lack of interoperability. Most companies have lots of different equipment for historical reasons, but it tends to mean the first thing you need is an interoperability lab.”

Titterington estimates you can easily spend 10 percent of a SAN-related budget on trials and tests.

“NAS is a cheaper and easier option, but SANs offer better security,” says Alvin Ong, regional sales manager for Maxtor.

Not everyone is convinced. “The inbuilt security mechanisms for SANs are pretty crude,” says Titterington. “It’s supposed to limit via server zones, but it’s alleged that even that can be breached.”

“What we are seeing now is a convergence of the two with NAS heads built upon SAN backbones, enabling enterprises to service all areas of the business with appropriate technology,” says Elisha.

Whatever the network architecture, new standards for high-speed data transfer such as iSCSI and Serial ATA are essential to make real-time access to high-volume storage a reality. Titterington believes the technology that we need most seriously is iSCSI—Gigabit Ethernet is only the precursor. But both options may need further enhancement.

“iSCSI is still very immature,” says Ray Morcos, systems engineer at SGI.

“Serial ATA on its own is nothing special,” says Bert Noah, enterprise solutions group director for Acer. “The development of interoperability for serial ATA will prove its success.”

At the moment, the most reliable interconnect remains Fibre Channel. Its high speed has been matched by a high price, but that is changing. “The price gap between SCSI and Fibre Channel has almost disappeared,” said Noah, predicting the two technologies will be equal in cost by year’s end.

Yet arguments over technology shouldn’t blind IT managers to the bigger question of business processes. “It doesn’t matter how you connect into it. The big bang for buck comes when you can get a single way to backup,” says Gold.

Advertisement

Talkback 0 comments

Latest Videos

Sponsored content

Power Centre - Content from our premier sponsors

Blogs

  • Suzanne Tindal IT: Govt's cost-cutting bitch
    The government needs to stop looking at IT as a necessary evil or the place to remove costs when the Treasurer comes calling.
  • Array Can complaints on mobile content be cut?
    On 1 July this year the new Mobile Premium Services Code was introduced. It sounds like it's had a good impact, but is it enough?
  • Array NZ farmers: Bleating about broadband
    As we know, farmers are such bleaters. They bleat as much as the four-legged woolly things in their paddocks. If it's not the weather, it's the strength of the dollar! Nothing is ever right. Likewise with rural broadband.
  • More blogs »

Tags

Back to top

Featured