IBM TS7650G PROTECTIER DEDUPLICATION GATEWAY Overview - page 9
9 of 11
P R O D U C T P R O F I L E
Copyright The TANEJA Group, Inc. 2008. All Rights Reserved
87 Elm Street, Suite 900 Hopkinton, MA 01748 Tel: 508-435-5040 Fax: 508-435-1530 www.tanejagroup.com
throughput as opposed to throughput that is
aggregated
across
several
independent
systems.
The introduction of clustering technology
has important implications in the areas of
performance and high availability. As
mentioned above, it allows IBM to increase
their in-line, single node performance lead in
the industry even further. Very high single
system throughput is most important when
customers have newer, higher performance
FC interfaces between the backup servers
and the VTL – just what you’d expect in the
large enterprise environments at which IBM
is targeting the TS7650G.
Availability is another extremely important
consideration
in
these
types
of
environments. In two node configurations, a
single node can fail and the remaining node
will immediately begin servicing the entire
workload, although the overall throughput of
the configuration will drop to that of a single
node. The failed node can be replaced on-
line and re-integrated into the cluster
without having to disrupt the backup
applications that are writing to the VTL.
Clustering also gives customers additional
flexibility in performing maintenance and
upgrades to cluster nodes, as well as
gracefully expanding cluster size in the
future as larger node counts are supported.
The TS7650G clustering technology supports
both improved performance and availability,
not just improved availability.
Evaluating the IBM TS7650G
How well does the TS7650G perform against
the criteria we identified earlier for
evaluating SCO VTL solutions (performance,
scalability, availability, reliability, solution
maturity)?
Performance. We’ve already reviewed the
TS7650G’s industry-leading in-line, single
node
and
single
system
performance
numbers, showing how that is directly
related to IBM’s patent-pending HyperFactor
de-duplication technology. The highly
efficient index design of HyperFactor allows
it to scale up to 1PB of base capacity without
impacting
indexing
performance,
a
considerable
problem
for
competitive
alternatives that are based on hashing or
content-aware algorithms. IBM’s roadmap
includes expanding the solution to a higher
number of nodes over time, which will offer
large enterprises a non-disruptive, long-term
growth
path
to
higher
performance.
Competing
vendors
may
offer
higher
aggregate throughput today, but single
system throughput is the operative number
for the enterprise data center. What is clear
is that the TS7650G supports the industry’s
highest in-line, single system throughput
performance for a SCO VTL today by a wide
margin.
Scalability. The data growth rates that
most large enterprises are experiencing today
mean that most will be managing at least
hundreds of terabytes of secondary data in
the near future. With ProtecTIER’s ability to
support up to 1PB of raw capacity, the
TS7650G can support multiple petabytes of
usable capacity, depending on the achieved
capacity optimization ratios across the
relevant workloads. Hash-based and
content-aware de-duplication algorithms do
not even come close to the scalability of