ur search engine crawled the Web.
The web site itself may have changed. You can check the current page or check for previous versions at the Internet Archive. Yahoo! is not affiliated with the authors of this page or responsible for its content.
UP
Copyright © 2003 by InfoWorld Media Group Inc., a subsidiary of IDG Communications Inc. Posted from InfoWorld, 501 Second Street, San Francisco, CA 94107.
#1-1588111 Managed by Reprint Management Services, (717) 399-1900. To purchase reprints online, visit www.reprintbuyer.com.
GET TECHNOLOGY RIGHT
September 15, 2003
b
ISSUE 36
UP
storage technology marches on. SANs are
faster, easier to scale, higher performing, and
more robust than ever individually. However,
the challenge of integrating multiple SANs
remains vexing. Although integrated reporting
and alerting is easy, its nearly impossible to
truly merge multiple vendors storage networks
without using invasive storage virtualization
servers that add their own layer of complexity
to the network.
That sums up the results of SAN management
tests performed by InfoWorld at the Advanced
BY ALAN ZEICHICK
SYNCHRONIZATION
EMC, HP, IBM, and Fujitsu Softek square off
in our SAN integration challenge
STEPPING
Sing
le Co
py
Use
Only Network Computing Lab at the University of
Hawaii in July. Immediately following Test
Center Senior Analyst Mario Apicellas hands-
on test of the latest SAN hardware solutions
from EMC, Hewlett-Packard, and IBM, the
integration portion of the test began
(
infoworld.com/240
). We asked the three
hardware vendors to merge all three SANs into
a single SAN. We also invited Fujitsu Softek,
as a hardware-agnostic storage software
provider, to take part in that challenge.
Challenge was the operative word in this
test, partially due to the complexity of the
task of merging heterogeneous SANs and
partially due to unexpected difficulties with
the test plan. Our goal was for the vendors to
offer as much functionality as they could, in
terms of unifying both the administration of
multiple SANs into one console, and then be
able to integrate storage resources between
the SANs.
True integration proved impossible for sev-
eral reasons. One was the newness of some of
the test hardware. The vendors hadnt seen
some of their competitors latest arrays yet and
didnt have drivers to support the hardware. In
other cases, vendors were loath to grant their
competitors unrestricted access to their cut-
ting-edge SAN hardware.
Further, our test plan did not stipulate a
known state that the network would be
left in, so that each vendor actually faced a
unique integration challenge of unraveling its
competitors products before attempting to
implement its own solution. Thus, in
acknowledgement of the lack of parity in the
test environment, weve chosen to forego scor-
ing this test, substituting instead a checklist
of features and functions demonstrated by the
vendors.
But the fact remains that SAN technology
hasnt evolved to the point where true integra-
tion is achievable, even under perfect condi-
tions (see Integration-Induced Frustration,
page 4).
The results: The software-based manage-
ment tools that EMC and HP provided could-
nt take full control of competing SAN hard-
ware, even when those SANs were integrated
at the fabric level by tying together the FC
(Fibre Channel) switches. They did offer uni-
fied administration of multiple SANs, which is
helpful, but not the panacea we were seeking.
By contrast, Fujitsu Softek and IBM offered
much more complete integration by inserting
their virtualization servers into the fabric and by
flowing all traffic through them (see Virtual-
ization Reality, page 5). Although Fujitsu
Softek provided the highest levels of integration
in our test, even it couldnt provide full access
to, and full control of, everything on the com-
bined SAN. Plus, the virtualization solutions
added another layer of software hierarchy to the
storage network and created a potential perfor-
mance bottleneck and single point of failure.
EMC and Hewlett-Packard, the two vendors
that supplied software-based storage manage-
ment solutions, werent able to completely inte-
grate the storage networks. In fact, they couldnt
get close: Their offerings were limited to pro-
viding unified administration of the different
SANs.
Thats not to say that those software tools
wouldnt be valuable; to the contrary, theyre
perfectly suited for many day-to-day operations
such as resource discovery, status reports, and
alerting in case of faults. Both perform
admirably, however, EMC has the definite edge
in terms of device support, functionality, and
toolset integration.
The Virtualized Solutions
The storage integration systems presented by
Fujitsu Softek and IBM are ostensibly similar,
each with a software console that presented a
unified view of the combined network, and a
virtualization server that went a step beyond
to consolidate all storage assets into a single
pool actually, into a single, large, virtual
drive that could be partitioned and allocated
to servers as needed. However, the solutions
differed in implementation.
IBMs software was focused on compatibil-
ity with IBMs storage arrays, and its Volume
Controller virtualization server was imple-
mented as a hardware-based appliance. By
contrast, Fujitsu Softeks software is designed
for heterogeneous environments, and its Stor-
age Provisioner virtualization solution, a soft-
ware application, can be installed on standard
x86 servers.
Based on the tests, we believe that Fujitsu
Softek has the most robust storage-networking
interoperability solution and a more intuitive
user interface. IBM ran a close second; IBMs
Volume Controller virtualization appliance was
only released a few weeks before the test, and
we expect it will swiftly evolve into a stronger
offering.
I N F O W O R L D . C O M 0 9 . 1 5 . 0 3
SAN Integration Scorecard
We assessed four
SAN management solutions. Heres the final tally.
FUJITSU SOFTEK
2
Yes
Yes
Yes
Yes
No
Yes
Yes
Yes
Yes
Yes
EMC
1
Yes
Yes
Yes
No
Yes
No
Yes
Yes
No
No
Assign storage and manage devices on one SAN
Unified management console
Autodiscover and map storage resources
Ability to broadly control multivendor hardware
Wizards available for administrative tasks
Tools for automating storage management
Robust role-based administration and access control
Policy-based storage automation
Move/span volumes across multiple SANs
Consolidate free space across integrated SAN
Note:
1
ControlCenter 5.1.2;
2
SANView 5.1, Storage Provisioner, and Storage Manager Server;
3
OpenView Storage Area Manager 3.1;
4
Tivoli
TotalStorage SAN Volume Controller, SAN Manager, and Storage Resource Manager
HP
3
Yes
No
Yes
No
No
No
No
No
No
No
IBM (TIVOLI)
4
Yes
No
Yes
No
No
Yes
No
Yes
No
Yes
Sing
le Co
py
Use
Only Fujitsu Softek
Fujitsu Softek started with a disadvantage
going into these tests. As the fourth and final
vendor, following the hardware manufactur-
ers, the entire storage network was in consid-
erable disarray. Nevertheless, the companys
SANView 5.1 SAN-management software
was able to perform discovery, analysis,
administration, and reporting across the entire
storage network with ease. But despite assis-
tance from Apicella, who had administrative
access to the EMC, HP, and IBM systems, the
Fujitsu Softek engineers had difficulty taking
full control of some storage systems on the
combined FC fabric, particularly the IBM gear.
Still, that didnt stop the plucky Fujitsu
Softek team. Concerned about exactly this
potential situation, they shipped their own
low-end disk array to the test lab, installed it
onto EMCs switch, and demonstrated their
storage virtualization solution on it as well as
on the EMC and HP portions of the test net-
work. This was enough to demonstrate not
only that their system works, but that both the
SANView software and Storage Provisioner
2.1 virtualizing server are more cross-platform
friendly than the other solutions used in this
test.
I was impressed by Storage Provisioners
flexibility. It can be deployed onto x86 servers
of any size; installed in very redundant
arrangements consisting of two, three, or
more servers; and configured to have a vary-
ing number of FC host bus adapters config-
ured either to virtualize servers or storage.
Considering that some networks may have
more servers than storage arrays and oth-
ers may have more storage than servers
the solutions extreme configurability is
noteworthy.
The user interface for Storage Provisioner
server is excellent. So, too, is the interface for
SANView SAN-management software, which
acts as the master control console for the inte-
grated SAN. Even without virtualization, SAN-
View goes far in unifying the administration of
the three storage networks, by providing con-
figuration information on the entire network
from a data repository in Softek Storage
Manager.
Unlike the other offerings, it doesnt play
favorites; theres no built-in bias toward a
specific set of hardware produc