problem, all workers had to log off of the email server while someone moved 

emails from the queue file by file. 

Further, the surge of disaster victim registrations resulting from the Florida 

hurricanes overloaded NEMIS' main server, pushing the system beyond its 

limit.  Originally designed to handle a maximum of 20,000 disaster victim 

registrations a day, during a four month period from August to December 

2004, NEMIS registrations far exceeded these limits during peak periods, 

reaching over 40,000 on some days.  The total number of disaster victim 

registrations processed during the four month period of the Florida hurricanes 

was 1,745,183.  The volume of transactions and the number of personnel 

managing these registrations significantly slowed down the system or made it 

unavailable for use during peak operations.   

Although EP&R CIO staff worked to keep the system operational by 

increasing system memory, NEMIS' main server became overloaded, the 

system froze, and unplanned system restarts were necessary.  Users were 

unable to perform their jobs in the system and consequently reverted to paper 

based methods.  When NEMIS' main server went down, approximately 2,000 

IT users were kicked out of the system for as long as 20 to 30 minutes at a 

time.  FEMA personnel accepting victim registrations had to record the 

information manually and wait to register the victims in NEMIS when the 

system was functional again.  Additionally, FEMA personnel lacked the up 

to date NEMIS information needed to answer disaster victim inquiries when 

they called the National Processing Service Center for assistance.   

As part of our review, we requested system performance reports from the 

EP&R CIO's office to determine how the systems performed during the 

hurricane response.  However, the CIO office did not have a standard process 

in place to produce system performance reports.  Instead, it had to complete a 

manual analysis of raw data to provide performance data for only one of 

NEMIS' key servers.  The CIO office stated that it would take several months 

to supply performance information for all of the other servers.  Although the 

CIO office can monitor central processing unit and hard drive space 

availability on a real time basis, it does not have a tool that can show system 

performance over time.  Without such a tool, it is difficult for FEMA to 

identify system performance problems and take corrective actions to address 

them. 

Given the problems experienced during the 2004 hurricane response and 

recovery season, a number of FEMA officials expressed concern about not 

only NEMIS' current capabilities, but also its capacity to support future 

Emergency Preparedness and Response Could Better Integrate Information Technology  

with Incident Response and Recovery 

Page 24