| Data storage and its management has always
been a nightmare and life does not seem to be getting any easier as the range of storage
systems grows. Not so long ago storage management was a relatively simple affair and
everyone knew their grandfather, father and son routine by heart. Compared to disk, tape
was a much cheaper, more reliable, long-term storage option and capacities grew more or
less in keeping with disk sizes. Finding time to do the backups may have been hell but the
principals were easy to grasp. |
Did you know?
The Data storage Market is expected to grow to 87 billion $ in year 2000, a rise of 20
billion $ from 98.
IDC says DLT market will grow by 30-40% each year for the next 3-4 years. Source: Quantum
| ATL |
|
New
technologies, new issues
Suddenly, we've seen the advent of laser storage technologies, massively higher capacity
disk drives and the accompanying technologies of RAID, clustering and Storage Area
Networks (SANs). As if this was not enough to handle, there are new issues to consider:
the intranet/e-commerce data boom, tape capacities trailing in the wake of the new disk
systems, the growing need for replication of data stores. Even the mechanics of tape
backing-up has become more complex with Hierarchical Storage Management (HSM) lurching out
from the shadowy world it initially inhabited.
Chris Boorman, marketing director at Veritas, summed up the current situation. "We've
become addicted to our computers. We can no longer do business without them. The amount of
data that we're storing is going through the roof. These machines are not infallible so we
have to backup the data but we also have to manage it more effectively than we did before.
Backup is just one component of storage management, managing disks and tapes, managing
where and how the information is stored, how we access it and ensuring that we can always
access it are all part of this picture."
Now, we have a new phrase to deal with, integrated storage management. According to
Boorman: "Our definition of integrated storage management is being able to manage the
entire storage environment and integrating management activities in such a way that you
gain distinct business benefits." The bottom line reads: the efficient management and
protection of digital information and applications wherever and however they may be
stored. A key technology in the future will be the Storage Area Network (SAN) with its
parallel Fibre Channel network enabling data to be moved where it is most needed or for
replication or back-up storage.
Back it up
Backing-up data has always been the preserve of creatures of the night (or the weekend at
least). The extra traffic generated by the housekeeping activities often reached such
volumes that it could not be undertaken during business hours. With the gradual
encroachment of business operations reaching the round-the-clock operations (24x7) levels
we see within corporates today, data management has been squeezed into ever smaller time
windows. This has brought about an incredible degree of efficiency in the way data is
managed and much of this will underpin the new integrated storage management scene.
Boorman explained, "Traditionally, backup software has simply taken data off disk in
a relatively coarse manner. For example, if a file changes I can do a full backup and take
copies of all the files on disk or, more usually, do an incremental backup of only those
files that have changed. In today's computing environment where files are getting bigger
and bigger identifying a file that has changed and simply backing the whole thing up is a
ridiculous way of operating."
What the leading back-up companies have done is to build intelligence into their software
that will examine the changed file and only backup the block that has changed. "In
large database environments," Boorman said, "this block level incremental backup
concept is quite profound because you may have a file that is gigabytes in size but has
only had a few blocks changed."
New possibilities
Another approach is snapshot technology where a secondary server is used to take a virtual
copy of its partner's resources. It is a form of fault tolerant clustering but instead of
the standby server being redundant it is linked to backup hardware. Chris Lentz, European
marketing manager responsible for Legato's high availability division, explained that
SnapShot Server was developed by his former company, Vinca, which was recently acquired by
Legato. "By providing a virtual copy of the primary server, SnapShot Server provides
a way for customers to take a backup of a defined volume without shutting down the real
volume. The system was developed for Novell NetWare but is now also available on NT."
The product takes its place in the company's latest initiative, Legato Continuum. This is
the broader picture that shows Legato, like Veritas, has aspirations in the integrated
data management field. The aim is to provide a single management suite that is
multi-platform and intelligent and complies with open standards. Lentz described the
concept behind Legato's Continuum strategy: "Legato's concept of Information
Continuance has tape backup at one end, moves up through off-site replication, high
availability clustering, with more fault tolerant applications such as SAN architectures
at the top end."
It is the last category that is currently attracting industry attention as the rise of the
SAN opens new possibilities. Ray Rice, business manager with CMS Peripherals, said that
there are plenty of examples that show these standards are needed. "You can't be sure
that it will work if you buy company A's RAID system, B's tape drive and company C's fibre
switches. Currently, switched fibre fabrics are largely sold by companies who have
certified that A, B and C will work together. It reminds me of the early days of SCSI
products" he commented.
Several companies are pushing ahead with developing an industry standard based on the SCSI
Extended Copy specifications from the EMC-initiated FibreAlliance for the Storage
Networking Industry Association (SNIA). Other members of FibreAlliance include Computer
Associates, Hewlett-Packard, Compaq, Veritas and Legato. CA is the first to break cover
and has announced the Storage Area Network Integrated Technology Initiative (SANITI).
E-commerce has brought about a resurgence of interest in storage management
CA's SANITI, working within the company's Unicenter TNG Framework, promises to offer
single console management for and autodiscovery of Fibre Channel hubs, SAN switches, and
storage devices. Jim Callaghan, CA's product manager for information management products,
said, "E-commerce has brought about a resurgence of interest in storage management.
Because most of the companies in this area have come to value their data more highly than
their physical assets, it has highlighted the need for management to ensure that it's
secure and that they can get to it when they need it."
His comments are reflected by Donal Madden, Compaq's storage product manager, who said,
"Every customer I speak to is facing the same problems. Data is growing like a weed,
particularly in the NT environment so they're now backing up twice as much data as they
were a year ago. Added to this there are remote users ringing in at all times of the day
and night. So call centres are opening for longer and there's twice as much data to back
up and 70% less time to do it in."
In its way, Windows NT has added to the data management problems through its distributed
nature. The Distributed File System is helping users to come to terms with distributed
storage but there has been little relief for IS managers. Microsoft is well aware of the
burgeoning data problem and will go some way to address the problem with Windows 2000. It
is working primarily with Veritas (which acquired Seagate Software last year) to provide a
degree of hierarchical storage management (HSM) but there is scepticism over how far this
will address the issue. Rice from CMS commented, "I'm by no means a Windows 2000
expert but I think the HSM element will be like NT Backup which was Seagate's Backup Exec
stripped down. It will introduce the user to HSM but probably won't be something they
would continue to use. There'll still be plenty of room for ISVs (Independent Software
Vendors)."
Storage Area Networking
Andrew Cheeseman, a senior consultant at Microsoft, predicted, "Microsoft is putting
a stronger emphasis on backup and recovery. Storage management will provide a single
interface to volumes and the way volumes are mounted is going to be more like the
mainframe realm. Storage Area Networking is going into overdrive and Windows 2000 is
really going into these technologies. A lot of solution providers of SANs have added some
very standard interfaces into their boxes to allow you to use volume pools and everything
that goes along with SANs in a Windows 2000 environment. You can now build some very high
availability solutions using SANs - it's a good complement for Windows 2000."
Boorman at Veritas believes that SANs provide the answer to many of the current problems
surrounding data protection and availability. The fact that data storage can be shared
across all servers on the SAN means a change in emphasis in the backup and restore arena.
"SAN is going to enable you to provide high levels of availability within Windows
2000 or NT in a way you couldn't do before," he explained. "When a server or an
application falls over you can move processing across to a similar server and resume the
service."
In the realm of backing-up there will no longer be a need to divert server processing
power to pump data out to tape archives because the SAN will contain its own intelligent
transport. The introduction of this hostless backup regime also has the potential to move
data around faster and the added bonus of leaner block level backups means that online
systems can be constantly backed-up throughout the day with zero impact on the users. The
intelligence will also mean that applications, databases and documents can be treated
differently and this could mean the end of separate proprietary backup systems for
databases - another advantage of integration.
Addressing concerns
One issue that will have to be addressed within Windows 2000 is the unsociable NT 4.0
habit of grabbing for itself every bit of storage that it sees. Rice pointed out that on a
SAN this would be disastrous: "Unix has much better volume mapping. There are some
patches available for NT 4.0 but it is very much a patch. Windows 2000 has to come up with
something better."
The data stored on corporate servers and desktop systems may be well served by the SAN but
what about the growing mobile workforce? The rule in the past has either been to bring the
notebooks into the LAN environment for backing-up or to quickly transfer as little data as
possible. In some cases, users have to be allocated specific times to log-in to the
corporate network to avoid exaggerating the end of the day/ early morning network traffic
rush hour. With the bandwidth provided by Fibre Channel, these concerns will be eased but
remote users will need to have access to the applications and data stored on the user
network. It is possible, but not yet apparent, that a new breed of connection will enable
the user to forget about backing up because it will be done through the SAN while they are
logged onto the in-house network.
Fibre Channel promises to be the analgesic that will relieve the data management headache.
Even if there are still those out there who need convincing of the need for backups, the
Y2K threat will probably change their attitude. If systems fall over at the end of this
year, data backup may be the only way to ensure that no data is lost - assuming of course
that the backup system is Y2K proof. To be truly integrated, data management must work
everywhere, with any device. In short, backing up data should be a no-brainer. It should
be a background process that only becomes apparent when something goes seriously wrong. If
an application becomes corrupted and falls over, the management software should see that
the service and users are switched to another server. Meanwhile, the failed application
should be automatically rebuilt so that the server can be brought online as quickly as
possible once the cause of the crash has been ascertained and fixed. If a drive fails, a
hot swap should be replenished with data regardless of whether it is in a RAID array or
not. All of this on top of the background noise of constant backing up of data.
At the end of the last century wealthy families employed servants who had their own
quarters linked to rooms around the house through a back stairs arrangement. This allowed
them to move from place to place to do the housekeeping, unseen by the family. At the end
of this century, we are seeing the SAN providing a similar backstairs system to allow IS
managers to access their systems and allow them to do their data "housekeeping"
invisibly and in a more integrated and better managed environment.

|
|