[ Pobierz całość w formacie PDF ]
servers and their direct-attached storage devices onto a SAN-based
storage network infrastructure that also spans the two data centers.
The servers handle file and print services, Lotus Notes, and internal
application development, employing a combination of operating sys-
tems that include Microsoft Windows 2000, Windows NT 4.0, and
Novell NetWare 4.11 (now being phased out).
We wanted to build disaster recovery into applications and hard-
ware from the ground up, which requires an enormous amount of
infrastructure on the corporate level, says Michael Obiedzinski,
DTCC s vice president of IT. SANs were the only way to go.
Ensuring Redundancy
The infrastructure incorporates redundancy to address all the major
potential points of failure. The company installed three Windows
Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)
Copyright © 2004 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
Case Studies
Chapter 6
140
2000 server clusters and dual SAN fabrics that extend across the two
geographically dispersed data centers, which are linked via multi-
strand fiber-optic DWDM connections. Twelve Brocade Silkworm
2800 Fibre Channel switches manage 10 TB of disk storage on EMC
Clarion FC400 arrays.
If a server fails, its resources can be picked up by a surviving
server in the other city, explains Ed Alsberg, DTCC s IT operations
director. If a site or array fails, the surviving server is presented
with a mirrored copy of the data. So we cover darn near any disaster
scenario.
Computer Associates ArcserveIT running on a Microsoft Windows
2000 server manages the storage subsystems and tape libraries. All
backups to tape are written automatically to a Storagetek L700 tape
silo at each data center. Each application has a primary server on
one site and a secondary server on the other with two-way disk mir-
roring.
The SAN-based backup topology has been in production for 14
months and works beautifully, Alsberg notes.
Callisma, the storage systems integrator that helped DTCC design
its dual-SAN fabric, sees a growing number of its customers 75 per-
cent, in fact implementing similar configurations, says Jeffrey Birr,
national director of storage networking for the integrator. They ve
gotten more concerned about disaster recovery in the past few
months, Birr declares. They want to control things in-house rather
than outsource, and they now have the technology to do it.
Seeking Safety and Soundness
While DTCC, with a 2-year head start, is in the vanguard of
advanced backup implementations, Alsberg emphasizes that the
company is usually very conservative when it comes to implementing
innovative technologies. Safety and soundness are top priorities for
everything we do, he declares. We move very slowly and cautiously.
For example, DTCC has no immediate plans to move its core busi-
ness applications from IBM Z900 mainframes to open systems and
SANs. With applications handling some $23 trillion in customer
assets, We don t make jumps to new technology lightly, Alsberg says.
Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)
Copyright © 2004 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.
Case Studies
Case Studies
141
The same goes for the company s Sun Solaris three-tiered Web
platform, which provides customers with access to online financial
services. We re taking the approach that production customer-
facing applications belong on our EMC Symmetrix/SRDF/ESCON
storage infrastructure, Alsberg states. ESCON and SRDF are
tried and true technologies, while the SAN fabric is a new technol-
ogy to us.
On the other hand, DTCC is highly confident that its dual-SAN
fabric, which should be fully in place this quarter, will provide more
than adequate backup and disaster recovery for Lotus Notes, file and
print servers, and internal application development servers.
SANs are expected to provide several big advantages, such as
enabling DTCC to administer four times as much storage without
needing to expand its current core IT staff of four. Says Alsberg, You
get a single view of all the storage required for hundreds of servers.
You always know how much storage you have available for new
requests, and centralized backup is really important when you have
hundreds of servers across two data centers.
Storage administrators use Brocade s Fabricwatch to trap events,
such as a switch or port crashing, and send them to Hewlett-
Packard s OpenView enterprise management system. For perfor-
mance management, DTCC uses McData s SAN Navigator.
SANs also make it possible to allocate storage capacity among
servers with much more flexibility, Alsberg says. Today, you can only
buy EMC storage in 36 and 72 GB capacity. If a server only needs 10
GB, the rest is wasted. On a SAN, you can give one server 10 GB and
use the rest where it s needed.
According to Obiedzinski, moving from direct-attached to SAN-
[ Pobierz całość w formacie PDF ]