|
Health First, the largest healthcare provider in Brevard County
Florida, is located on the east coast of the state in hurricane
country, operating three hospitals with 800 beds that are served
by 6,000 employees. Health First also manages outpatient centers
and services, including: the county’s only trauma center;
home care; specialized programs for cancer, diabetes, heart, stroke,
and rehabilitative services. To promote internal efficiency, Health
First depends heavily on automation and thus needs IT systems availability
to ensure operational continuity. In 2002, Health First had approximately
16 terabytes of stored data.
Background
To protect this data, Health First had implemented
a traditional disaster recovery methods using tape backup. Backups
were made and sent daily to a media vaulting facility in Orlando
Florida. During a disaster, the vaulted tapes in Orlando would be
sent to a third party disaster recovery facility 1,200 miles away
in New Jersey, where duplicate servers, configured with key applications
resided. In the event of a disaster, Health First IT specialists
would be flown to New Jersey to recover data from the backup tapes,
bring up standby servers and applications, and connect the new site
to the Health First facilities using a T1 link. Although this widely
accepted, traditional form of disaster recovery had met their needs,
it became clear that in the future it would not. Within the last
3 years, Health First data had grown a staggering 300 percent. Health
First determined that the exponential growth of data that they had
witnessed was going to dramatically increase, and that scaling the
present recovery system would be too difficult and too expensive.
Objective
Hurricane logistics revealed that Health
First would have difficulties in deploying its IT staff during a
disaster to restore operations. Even if staff deployment was successful,
the minimum recovery time was expected to be 12-24 hours, based
on time for travel and restoring data from tape. Moreover, under
the best of conditions, the restored data would be one day old,
which would impair patient diagnosis, clinical documentation systems,
medical imaging stations, and physician order management. In addition,
only a few applications could be brought up within the needed recovery
time.
|  |
Using a T1 link to connect
client stations to the new data center, was also unacceptably slow.
Steve Shim, manager of Health First’s new disaster recovery
strategy explained, “The T1 line connecting our client stations
to the disaster recovery site simply doesn’t have the bandwidth
to address our needs, so we were forced to look for a new type of
local or regional disaster recovery strategy.”
The Solution
In searching for a new solution, Health
First developed new disaster recovery goals in the event of a hurricane-type
disaster. These were: 2 hours for 1-2 hospital failover; 2 days
for multi-hospital recovery; “best effort” for a 4-site
disaster. The new recovery strategy required delivery of continuous
IT services for three hospitals and one business center. Central
to this strategy was the idea of mirroring data regionally among
data center sites in real-time using a DWDM link, thereby reducing
recovery time from days to hours. Health First initially evaluated
array-based replication options, but soon found them economically
infeasible because all required duplicate sets of high-cost arrays
and software. Array-based replication was found to be neither instantaneous
nor automatic.
Other solutions such as copy services and NAS also fell
short. Host-based mirroring, the final alternative, was
eliminated because of the difficulty of managing mirroring
on multiple hosts, lack of scalability, and impact on server
performance. Fortunately, Health First’s Steve Shim
had heard of Vicom’s SAN data migration solution,
and he correctly guessed that Vicom might also have a SAN
mirroring capability. He said: “Our assumption proved
right: Vicom has a superior and unique product that solves
a big problem for Health First.” Using a purpose-built
appliance,Vicom SAN Mirroring provides identical copies
of data at all times and “instantaneous data recovery”
in the event of a disaster. The solution encapsulates two
logical units from different storage systems (or SANs) and
presents them to the host as a virtual, “unbreakable”
disk. Data written from the host is automatically and synchronously
sent to logical units via the Vicom Engine. With mirroring,
any failure to one of the units is shielded from the host,
and the Vicom Engine performs failover instantly and transparently.
|
|
|