Last August, Storage's first Purchasing Intentions Survey (see the August 2002 issue of Storage) predicted that 2003 would be a year of increased investment in storage by corporate IT departments, with the emphasis on storage area networks (SANs). That prediction was correct, according to our just-completed second Purchasing Intentions Survey. But while overall spending is up, we may finally be seeing evidence of a shift away from years of buying raw disk capacity toward a more even spending plan.
The survey of 515 storage managers and administrators (see "How we surveyed") showed surprisingly strong spending plans. Half of all respondents plan to increase spending in 2003, with 25% increasing by more than 10%. Only 22% plan to cut spending, with the remainder unsure (10%) or holding pat (18%).
|Selective new tech purchases|
Conventional wisdom says IT shops pass on new technologies in tough times. The contrarian's perspective is that only new technologies can help a company grow out of recession. Where did our respondents come down on this debate?
Somewhere in the middle.
We asked their plans for evaluating or implementing nine newer technologies: third-party snapshot/replication, SAN/NAS gateways, quota management, chargeback, iSCSI, virtual SAN, content-addressable storage, autoprovisioning and serial ATA. Some of these, of course, are not really new technologies, but ones that are getting renewed interest in the networked environment.
Three technologies--third-party snap replication, SAN/NAS gateways and quota management--have been or will be implemented by at least 20% of respondents. An additional 40% are or will be evaluating third-party snap replications, with 36% and 38% looking into SAN/NAS and quotas.
Two other technologies--Serial ATA (SATA) and iSCSI--are also high on the evaluation list, at 28% and 30%. Unlike the other three technologies with high interest, these are largely unavailable now, so they had single digit implementation scores.
As for the rest, two technologies highly associated with a single vendor--virtual SAN with Cisco and content-addressable storage with EMC--showed little implementation, some interest (30% and 32% evaluation scores), but high negatives (as they say in politics). In the "definitely not implementing" and "no plans" categories, they scored 69% and 66% respectively.
Without question, the big loser among the new technologies we queried users on was auto provisioning. With a majority of users (69%) indicating no plans or definite disinterest, and only 27% even considering an evaluation, this technology would seem to have a ways to go in users' minds.
Central to users' spending plans is the continuing push into networked storage. Nearly 60% of respondents said more than half of their storage spending would be for networked storage, up from 54% last August. The lion's share will go toward SANs--50% said SANs would be their primary expenditure for new disk, up by 3% from last August. The big loser was network-attached storage (NAS), which dropped from 20% to 11%. The growing primacy of SANs was even more evident for new applications, where 58% of respondents tabbed them as primary storage.
What's driving this push? Consolidation. That's what 42% of respondents cited as their primary reason for deploying network storage, with backup/disaster recovery a distant second at 26%. Last summer, the two were at parity as main drivers. Both surveys showed that databases continue to drive SAN data growth, with e-mail and files playing an important, but secondary role.
But all of this SAN emphasis won't mean unbridled capacity growth, according to respondents. While last August's survey showed most respondents buying more than 10TB (61%), this year's shows the opposite--75% will buy 10 or fewer terabytes. That dovetails with where users said they were last August, were the emphasis was on creating new SANs, while this year the focus is on adding devices to the SAN. The capacity binge of the last several years appears to be over.
The state of SANs that emerges from the survey suggests that industry hype is well ahead of reality: most users are either running large director-class switches or small SAN islands. Only 20% described themselves as having a single fabric made up of large core switches and smaller edge switches. However, 24% indicated that was their direction for 2003.
Those who have opted for large, director-class switches (20%) are sticking with that approach for 2003. But they aren't typical: 78% of respondents are buying switches with 32 or fewer ports. A similarly large number are buying 10 or fewer switches. The most popular plan for 2003 is to deploy islands of small switches (29%), so SANs are still at a modest stage at most companies.
No slowdown in tape
While backup and disaster recovery slipped in urgency since last summer, tape is still high on the agenda. A full 60% of respondents are increasing spending on disaster recovery, with 40% increasing spending on off-site tape, online backup services and backup software. Fewer than 10% of respondents were decreasing spending on major activities associated with backup and disaster recovery.
At the same time, there doesn't appear to be a radical shift in how people are accomplishing backup or disaster recovery. Spending for online backup services won't change for 56% of respondents. Only 39% were increasing spending on remote mirroring, for which 43% said they would have no spending on at all.
Tape backup and off-site storage--business as usual--seems to be the order of the day for many companies. Tape backup will increase at 55% of companies, and only decrease at 12%. Those increasing their tape use cited increasing numbers of applications or users buying into the backup plan as their major motivation (43%), with business continuance/disaster recovery at 32%. Of those who are decreasing spending on tape, 54% pointed to disk-based backup as the main reason.
If tape is the first option for backup and disaster recovery, users are apparently taking an incremental approach. Respondents were in a buying mood for tape drives--64% will be buying them, although most (51%) will only buy one to 10 drives. As for libraries and autoloaders, most respondents weren't planning on buying any (52% and 61% respectively).
As for the tape format horse race, don't look for any big shifts next year. LTO and DLT/SDLT remain virtually tied as the primary format, with a third of the market each. The 9840 format was a distant third at 9%.
|Is this the year of IP storage?|
Last summer's survey showed strong initial interest in IP storage, particularly iSCSI. More than 20% indicated that one or another IP protocol would be their main storage protocol in 2003, with iSCSI hitting 14%.
That may have been an expression of interest more than a statement of intent. This year, IP protocols got the nod as primary protocol from 11%, with iSCSI at 8%. Real interest in IP storage is driving purchases of Fibre Channel to IP bridges by 30% of users, and iSCSI to SCSI bridges by 18% of respondents. In fact, 3% of respondents intend to purchase more than five iSCSI bridges.
Nonetheless, 25% of respondents said they would deploy IP storage switches in 2003. A realistic 3% of respondents said they had already implemented iSCSI, with another 3% saying they definitely will implement it. But 6% were in the process of evaluating it, while 28% said they would evaluate iSCSI.
Yet iSCSI is not everyone's cup of tea: 10% are definitely not implementing it and 49% had no plans to implement it.
All in all, it would appear that 2003 is the year that iSCSI gets on the radar screen, but maybe not on the shop floor just yet.
Caution on management software
With capacity spending not as prevalent as in the past, storage managers might be expected to be putting more emphasis on managing their environments through software. But the picture is muddier.
Nearly half of respondents (49%) are increasing spending on storage management software. But 21% are spending nothing, and 26% are holding the line. For those buying, there's no doubt as to why: 42% listed the need to manage more storage with the same staff, with 10% more trying to manage existing storage with the same staff.
Perhaps the most interesting data from the survey on software was an apparent shift backward to more basic software. In August, the emphasis was on packages to manage multiple elements, with some interest in integrating into higher level suites, such as Hewlett-Packard OpenView or Computer Associates Unicenter. There was little interest (18%) in element managers. But this time around, interest in element managers and integration into higher-level suites switched rank, with multielement managers still holding the high ground. Now that they're deeper into it, are storage managers realizing the need to put in software controls at the most basic levels before moving on to more ambitious plans?
We'll repeat the survey this summer during the early phases of your 2004 corporate budget planning. Perhaps that survey will tell us how storage managers' plans to grow in 2003 survive a poor business climate, and whether the tendency to get the house in order before massive expansion can hold out against the realities of explosive data growth.